Prévia do material em texto
<p>Contents</p><p>Introduction</p><p>Chapter 1: Appraising Theory</p><p>Outlining the Testing Program</p><p>Simulation Example</p><p>Overview of Ground Rules for Testing</p><p>Selective Testing</p><p>Chapter 2: Index Regression</p><p>What’s Relative?</p><p>The Evolutionary Process: Down from the Trees</p><p>A Concept is Born: Adaptation of Indexes to</p><p>Price Action</p><p>Creating the Channels</p><p>Theoretical ways to Trade Price Action Using</p><p>Channels</p><p>Establishing a Benchmark – Relative Price</p><p>Channel</p><p>Conclusions to Draw at this Early Stage</p><p>Coding for Chapter 2</p><p>Chapter 3: The Emergence of Relative Adaptive</p><p>Analysis</p><p>Converting an Open Oscillator into a Range-</p><p>bound Index</p><p>Internal Strength Index</p><p>Applying the ISI to Price Action</p><p>Establishing a Benchmark – Internal Strength</p><p>Channel</p><p>Coding for Chapter 3</p><p>Chapter 4: The Final Frontier (Well, Almost)</p><p>Directional Movement</p><p>Establishing the Benchmark</p><p>Conclusions so Far</p><p>Coding for Chapter 4</p><p>Chapter 5: All Things Relative</p><p>Relative Strength Comparison</p><p>Comparative Strength Index</p><p>Comparative Strength Channel</p><p>Establishing a Benchmark</p><p>Coding for Chapter 5</p><p>Chapter 6: Mingle with the Crowd</p><p>Volume Bias Index</p><p>Volume Bias Channel</p><p>Establishing a Benchmark</p><p>The End of the Benchmarks</p><p>Coding for Chapter 6</p><p>Chapter 7: Compatibility</p><p>Relative Relationships −14 Periods</p><p>Stand-alone Confirmation</p><p>Beyond the Benchmark – the Relative Price</p><p>Channel at 34 Periods</p><p>Beyond the Benchmark – Internal Strength</p><p>Channel at 34 Periods</p><p>Beyond the Benchmark — Directional Movement</p><p>Channel at 34 Periods</p><p>Conclusions So Far</p><p>Beyond the Benchmark – Volume Bias Channel</p><p>at 34 Periods</p><p>Relative Analysis and Compatibility</p><p>Chapter 8: Confirmation</p><p>Channels Combined with Initial Highs</p><p>Initial Highs</p><p>Establishing a Benchmark</p><p>Beyond the Benchmark — Adding Confirmation</p><p>Beyond the Benchmark — 55-period Relative</p><p>Channel as Confirmation</p><p>The End Result</p><p>Using the Lessons so Far</p><p>Initial High and Daily Comparative Strength</p><p>Channel</p><p>Initial High and Weekly Comparative Strength</p><p>Channel</p><p>Coding for Chapter 8</p><p>Chapter 9: Bilateral Relationships</p><p>Crossing the Neutral Zone — Bullish</p><p>Long-Term Bullish Behaviour</p><p>Crossing the Neutral Zone — Bearish</p><p>Understanding Complex Behaviour</p><p>Patterns</p><p>Chapter 10: Relative Trailing Stops – Part One</p><p>Relative Trailing Stops</p><p>Choosing an Effective Trailing Stop</p><p>Initial High/Mean Close Trailing Stop</p><p>Benchmark — Daily</p><p>ATR Trailing Stop — Daily</p><p>Relative Percentage Trailing Stop — Daily</p><p>Bear Range Trailing Stop — Daily</p><p>Arbitrary Percentage Stop – Daily</p><p>Conclusion – Daily</p><p>Bear Range Trailing Stop – Personal Application</p><p>Chapter 11: Relative Stops – Part Two</p><p>Trailing Stops on Weekly</p><p>Mean Close Trailing Stop Benchmark – Weekly</p><p>ATR Trailing Stop – Weekly</p><p>Relative Percentage Trailing Stop – Weekly</p><p>Bear Range Stop – Weekly</p><p>10 Per Cent Trailing Stop – Weekly</p><p>Summaries of Results</p><p>Conclusions</p><p>Chapter 12: Trailingstops and Price Channels</p><p>Relative Price Channel and ATR Trailing Stop</p><p>A Close Below the Exit Signal</p><p>Overbought Exit on Sustainable Price Action</p><p>Overbought Exit on Unsustainable Price Action</p><p>Further Research</p><p>Coding for Chapters 10 to 12</p><p>Chapter 13: Taking the Next Step</p><p>Index</p><p>First published 2006 by Wrightbooks</p><p>an imprint of John Wiley & Sons Australia, Ltd</p><p>42 McDougall Street, Milton Qld 4064</p><p>Offices also in Sydney and Melbourne</p><p>Typeset in Adobe Garamond 12/14.4 pt</p><p>© Leon Wilson 2006</p><p>The moral rights of the author have been asserted</p><p>National Library of Australia</p><p>Cataloguing-in-Publication data:</p><p>Wilson, Leon, 1963-.</p><p>Breakthrough trading: revolutionary thinking in relative analysis.</p><p>Includes index.</p><p>ISBN 0 7314 04424.</p><p>1. Stockholders - Australia. 2. Stock exchanges - Australia. 3.</p><p>Investments - Australia.</p><p>I. Title.</p><p>332.63220994</p><p>All rights reserved. Except as permitted under the Australian Copyright Act</p><p>1968 (for example, a fair dealing for the purposes of study, research, criticism</p><p>or review), no part of this book may be reproduced, stored in a retrieval</p><p>system, communicated or transmitted in any form or by any means without</p><p>prior written permission. All inquiries should be made to the publisher at the</p><p>address above.</p><p>COver design by Rob Cowpe</p><p>Disclaimer</p><p>The material in this publication is of the nature of general comment only, and</p><p>does not represent professional advice. It is not intended to provide specific</p><p>guidance for particular circumstances and it should not be relied on as the</p><p>basis for any decision to take action or not take action on any matter which it</p><p>cOvers. Readers should obtain professional advice where appropriate, before</p><p>making any such decision. To the maximum extent permitted by law, the</p><p>author and publisher disclaim all responsibility and liability to any person,</p><p>arising directly or indirectly from any person taking or not taking action</p><p>based upon the information in this publication.</p><p>Note</p><p>I have no formal business relationship with Compuvision Pty Ltd, the</p><p>developers of TradeSim, nor do I receive any form of payment or</p><p>commissions from Compuvision for mentioning TradeSim, either directly or</p><p>indirectly. As a private trader I mention those programs that I find of benefit</p><p>in real time. The TradeSim product produced by US Company Trade Lab</p><p>Strategies is not the same product as used by me for research and</p><p>development purposes or discussed throughout this book. Readers interested</p><p>in evaluating TradeSim as discussed here and determining whether it is</p><p>appropriate for their trading needs should log onto Compuvision’s website</p><p>for further details <www.compuvision.com.au>.</p><p>http://www.compuvision.com.au</p><p>Acknowledgements</p><p>Before I start, a special thank you must go to my beloved wife, Louise, for</p><p>being there as I compiled this latest book. Her unquestioning help and</p><p>ongoing support are priceless, as they allow for projects such as this book to</p><p>materialise. Without Louise there would be no book and trading would be</p><p>inherently more difficult.</p><p>Often those behind the scenes receive little or no acknowledgement for</p><p>their input so, while I may not name you personally, you know who you are.</p><p>A sincere thank you to those people who have contributed to making this</p><p>book happen.</p><p>John Hay happened to be just the best and most loyal mate that anyone</p><p>could ever wish to have and I looked upon him as my brother. His ability to</p><p>Overcome adversity which the majority of us would be unable to even</p><p>comprehend is an endless source of inspiration. Couple this with a most</p><p>mischievous sense of humour and life was never dull where John Boy was</p><p>involved. The two people with whom I was closest are not here today;</p><p>however, this is not so much sad as concerning. I’m sure that John now sits</p><p>with my brother Ross, plotting some devious scheme in order to derail my</p><p>well-meant plans — which they will find most humorous at my expense. Is it</p><p>too late to apologise for all my practical jokes?</p><p>Something to think about</p><p>Today we are so busy making a living that we forget that it is a privilege to</p><p>live. Think of where we originated, the odds of it happening and, ultimately,</p><p>the odds of survival today. Take the time to smell the roses along the road of</p><p>life and to enjoy each day as it unfolds — after all, we trade to live, not live</p><p>to trade.</p><p>Life unfolds with each rising of the sun; assumption leads us to believe</p><p>that we will be here tomorrow to see it</p><p>– Leon Wilson.</p><p>Introduction</p><p>Trading is a difficult occupation that is beyond the grasp of most, simply</p><p>because they lack the willingness to serve an apprenticeship — usually</p><p>fuelled by the inability to listen to others (a fault normally spelt ‘ego’). I have</p><p>always argued that traders all pay for their education when it comes to trading</p><p>— whether they like it or not. You can either sign up for a course offered by</p><p>institutions such as ACOFE (Australian College of Financial Education) and</p><p>others, or tackle the market head-on as I did. In my own defence, courses</p><p>were non-existent when I started out, so my options were limited. Entering</p><p>the market with the belief that you are pretty shrewd when it comes to buying</p><p>and selling while profiting in the process means you have set yourself on a</p><p>course for disaster, because this</p><p>at the time it would have caused many</p><p>traders some concern. Some will say that while price action later failed, the</p><p>concept worked; however, they are missing the point. A retrospective</p><p>evaluation has been allowed to have an influence. A divergence does not</p><p>change in appearance as price action evolves — it becomes assigned to the</p><p>history books in the same manner as it developed. It is always there for all to</p><p>see. However, patterns such as failure swings, which are only one of many</p><p>potential points in indicator development, can be absorbed into surrounding</p><p>indicator development as the trend evolves, meaning that what was a</p><p>significant point in trend development last week can be suddenly rendered</p><p>immaterial this week. Analysis techniques that are influenced by ensuing</p><p>trend development are of little use in real time as their relevance shifts in</p><p>accordance with price behaviour. With figure 2.3, two areas of significance</p><p>can be identified:</p><p>1 an extremely high RSI value that coincides with early trend</p><p>development</p><p>2 true divergence activity.</p><p>While it is difficult to confirm the appearance of failure swings until after the</p><p>fact, divergences are a representation of collective behaviour Over an</p><p>extended period of indicator development.</p><p>The bulk of conventional analysis usually involves the elements that I have</p><p>just cOvered — extreme values, divergences and failure swings. The problem</p><p>is that much of this analysis can become very subjective and is best</p><p>confirmed retrospectively. I wanted more from my confirmation process.</p><p>The evolutionary process: down from the</p><p>trees</p><p>While basic RSI application is suitable for identifying extreme points of trend</p><p>development, I wanted to visually quantify the relationship between bullish</p><p>and bearish activity. My initial idea was to redesign the conventional RSI</p><p>from index to oscillator format. This is a very straightforward process and</p><p>simply required the subtraction of 50 from the indicator value — knowing</p><p>that ‘50’ represents equal bullish and bearish activity within the conventional</p><p>index. By subtracting 50, neutral price action would produce a value of zero</p><p>while positive activity — that is, indicator values greater than 50 — would</p><p>produce a value greater than zero. Naturally, negative indicator values — that</p><p>is, values less than 50 — would produce a negative value. Hence, I</p><p>effectively turned the index into an oscillator. While the representation is</p><p>identical to the conventional index format, the visual representation can take</p><p>on another perspective without distorting the original value — that is, it can</p><p>visually define bullish and bearish internal strength.</p><p>I created the oscillator in order to assist with my own understanding of the</p><p>internal strength being experienced and to reveal the full value of this</p><p>strength — as traders, you should also think about what is in front of you and</p><p>how you may improve on it. If you enhance your understanding of what you</p><p>are being told, this will give you a better insight into what is really</p><p>happening. At the time, I failed to realise the world that was about to open up</p><p>simply by turning the RSI into an oscillator; however, it was this single step</p><p>that eventually led to the evolution of relative analysis.</p><p>In figure 2.4, I have applied both the conventional RSI and the relative</p><p>strength oscillator (RSO) to National Australia Bank (NAB) for direct</p><p>comparison. The Overbought region now becomes +20 while the Oversold</p><p>region translates to −20. These values represent the same regions identified</p><p>by the conventional index at 70 per cent and 30 per cent.</p><p>Figure 2.4: relative strength index and relative strength oscillator</p><p>(histogram)</p><p>Comparing the index to the oscillator highlights the following points:</p><p>The oscillator more clearly identifies the early signs of down trending</p><p>activity by allowing trendline placement to be applied Over a relatively</p><p>short period. If you were to focus solely on the oscillator format, it is</p><p>unlikely that you would continue to plot the trendline in the same</p><p>manner as I applied it to the more conventional index, especially during</p><p>2003 and 2004.</p><p>While the same indicator is applied using the same time periods, the</p><p>conventional index is far less decisive — especially throughout 2003</p><p>and 2004 — from a visual perspective. The identification and, more</p><p>importantly, the separation of bullish and bearish dominance now allows</p><p>you to scrutinise indicator activity with a degree of clarity previously</p><p>missing from analysis.</p><p>The early onset of continued bearish activity is identified using the</p><p>oscillator, while the conventional index identifies the upside break-out</p><p>Over the longer term decline. This highlights that combining both the</p><p>oscillator and the index formats has the potential to identify the same</p><p>information from two perspectives, as achieved here.</p><p>Oscillator values</p><p>While many traders will appreciate that 50 represents neutral activity with</p><p>indicators such as the RSI, its development can take on added significance</p><p>when this relationship is more clearly defined. I will not explore the</p><p>characteristics of the oscillator here in depth as it is the concept of an</p><p>alternative perspective to your approach that I am attempting to portray;</p><p>however, I will discuss oscillator values and what they represent. As already</p><p>discussed, the values produced by the RSI or RSO display the underlying bias</p><p>present within price action. While this can be consistent with the average</p><p>number of up or down days, this is not always the case. Having discussed the</p><p>basics, I will move much closer to the heart of what the values represent for</p><p>the more experienced trader.</p><p>Table 2.1 compares RSO values to RSI percentages. What these values</p><p>represent are as follows:</p><p>Table 2.1: relative oscillator values</p><p>Indicator value % Bias</p><p>50 100.0% (bullish)</p><p>40 80.0%</p><p>30 60.0%</p><p>20 40.0%</p><p>10 20.0%</p><p>0 0.0% (neutral)</p><p>−10 20.0%</p><p>−20 40.0%</p><p>−30 60.0%</p><p>−40 80.0%</p><p>−50 100.0% (bearish)</p><p>A value of 0 on the RSO indicates that the average range experienced by</p><p>the up days is comparable to the average range experienced by the down</p><p>days.</p><p>An indicator value of 20 reflects a 40 per cent bias in favour of the range</p><p>driving up days Over the range experienced by down days.</p><p>An indicator value of −20 reflects a 40 per cent bias in favour of the</p><p>range driving down days Over the range experienced by up days.</p><p>I will briefly cOver some of the crossOver rules for those who like to apply a</p><p>more conventional approach. The same principles apply for both index and</p><p>oscillator formats. The key when turning any index into an oscillator is</p><p>correctly identifying the point of neutrality, and the Overbought and Oversold</p><p>regions. With indicators such as the relative strength and the money flow</p><p>indices, this happens to be 50 per cent, 70 per cent and 30 per cent</p><p>respectively when used in a conventional manner.</p><p>The crossOver rules that can be applied are as follows:</p><p>An upward crossOver from the Oversold (−20) region is considered a</p><p>buy signal. (Once again, the RSO should be accompanied by price</p><p>action to confirm that it is not just gravitating back to a more neutral</p><p>value.)</p><p>A value greater than zero and rising suggests that a bullish bias exists</p><p>and that bias is increasing.</p><p>A value greater than zero and declining suggests that a bullish bias</p><p>exists but that bias is easing.</p><p>A value less than zero and decreasing further (moving away from zero)</p><p>suggests that a bearish bias exists and that bias is increasing.</p><p>A value less than zero but improving (moving toward zero) suggests that</p><p>a bearish bias exists but that bias is easing.</p><p>A downward crossOver from the Overbought (+20) region is considered</p><p>a sell signal. (Again, the RSO should be accompanied by price action to</p><p>minimise the possibility that it is just gravitating back to a more neutral</p><p>value.)</p><p>An upward crossOver from the Oversold region on below average or</p><p>low relative volume could be a potential consolidation in price action in</p><p>an existing downtrend.</p><p>A downward crossOver from the Overbought region on below average</p><p>or low relative volume could be</p>onto an equally efficient
confirmation tool, and without doubt one of my favourites, directional
movement. Most trading programs with coding capabilities similar to
BullCharts and MetaStock can accommodate the internal strength index. The
same programs will also include directional movement as a part of the
indicator menu, so it is logical to direct the focus toward these tools and this
style of charting software — although the concepts discussed can be
effectively applied to any non–price based indicator without the requirement
of major recoding. I will also build on previous discussions of indicator
values in The Next Step to Share Trading Success, as this directly relates to
the adaptation of indicators to price action. Later chapters will also look at
effectively combining indicators with entry and exit techniques.
The range in application is almost infinite and beyond the scope of what I
can cOver here in one book. In reality, I could write several books on the
topic and still not cOver relative channels comprehensively. Take the time to
learn and decipher the information that I have put before you and build upon
it to the best of your understanding. In some cases, your conclusions will
differ from mine; however, this does not imply that either is incorrect. They
are just different, relative to the strategy in place. The more I think, the more
I understand, the more I realise how little I know, the more my mind races
with questions, new ideas and concepts. I have barely scratched the surface of
adaptive relative analysis here and how I apply it to my everyday trading;
however, I feel that I have given you sufficient information to get started
without Overly confusing the issue. When reading through this book try to
understand the underlying concept and how it relates to your personal
understanding of price behaviour, rather than becoming tied down with minor
and potentially immaterial points of interest.
Finally, please remember that what you are reading are the findings and
observations of just one person. Where conventional analysis is concerned,
there is the collective opinion of many traders that has been accumulated and
documented Over literally thousands of hours of technical analysis, finally
culminating in the understanding and accepted practice of today.
This is not the case here. I have not read about this style of analysis
elsewhere. I have searched for similar publications in order to broaden my
own understanding, but to date my search has proved fruitless. All you have
is my personal observations drawn from everyday trading activities;
therefore, you need to approach this style of analysis with a very open and
investigative mind. It is simply impossible for me to identify and catalogue
all aspects of this approach relative to all situations and strategies, as much of
the sailing is done in uncharted waters. I can only relay to you my
observations from my perspective in relation to how I approach the market.
Once you start exploring the concept personally, I suggest that, as I do, you
keep a notebook handy.
By the time you finish reading this book I suspect that the majority of you
will never look at a 14-period indicator in the same light again. When
someone stands before you singing the praises of a 14-period relative strength
index — or, for that matter, of an arbitrary percentage trailing stop — you
will feel the urge to slink down in your seat and hope the topic will suddenly
change.
I wish you all the best and I hope that you find the content as I initially
intended — thought-provoking.
Cheers
Leon Wilson
March 2006
Chapter 1
Appraising theory
If you are happy to trade using accepted practice and have no desire to probe
the boundaries of technical analysis, there are plenty of solid technical trading
programs at your disposal that will accommodate a more conventional
approach. Many of these techniques have been tested Over time and proven
effective, so all that is required on your part is familiarisation. The weakest
link initially is you, as you come to terms with the strategy, tools and general
application in real time. However, the potential problem with such programs
Over the long term is that you will outgrow the software. Some of you may
say that it is unlikely you will want to move beyond your present software;
however, the continual learning curve that develops as you are trading and
that leads to a deeper understanding of the market eventually has most people
asking questions that basic and very rigid trading programs are simply unable
to answer. This does not imply that the product is suddenly inferior, just that
you are moving beyond its capabilities.
Over the course of this chapter, I will develop and test a basic trading
strategy, and discuss which factors within the testing results to focus on. The
discussion will highlight the factors that will then be used to compare the test
results of each new concept that is tested throughout the rest of the book.
Outlining the testing program
If you are going to develop strategies and continually probe the boundaries of
technical analysis, a quality testing program is essential in today’s trading
environment. For years such products simply did not exist, unless you were
prepared to purchase TradeStation. (MetaStock does include the system
tester; however, in my opinion it is a very rudimentary feature when
compared to specialised development software such as TradeSim.) In the era
before simulation and development software, research and development was
a very laborious task — to put it mildly. It involved many spreadsheets, real-
time trading and a very basic assessment process of historical data in order to
arrive at what was usually a questionable result. For example, I spent a
considerable period of time refining the parameters for my relative
percentage trailing stop. The best I could determine from this was that a
longer period of around 21 days was more appropriate than the more
accepted five periods applied to many trailing stops, and that 2.7 seemed like
an ideal multiplication factor. While this worked well in real time, by the
time everything was running smoothly the development process from start to
finish had taken almost two years. Then I discOvered TradeSim. I spent one
rainy afternoon reviewing my favourite trailing stop using TradeSim and,
would you believe it, the program identified that a 2.7×(21) relative
percentage trailing stop was ideal. What had taken me almost two years, was
achieved in one afternoon. While this might disappoint some, I was ecstatic
as it indicated that TradeSim had the genuine capacity to identify suitable
values relative to real-time application. Occasionally one program will stand
out as beneficial to your trading activities.
With regards to technical trading programs, there is an ever-increasing
selection of products all bidding for the trader’s dollar. Generally, they are
very similar in content and functionality. They will all have their own unique
features in one form or another and that one special indicator that
distinguishes them from their competitor; however, by and large, the
differentiation between products is usually minimal with many functions
once considered specialised now commonplace among most trading
packages. While some programs do have custom indicators, similar versions
seem to appear relatively quickly so any benefit from a proprietary indicator
is often limited.
TradeSim does not fit into this category. It is not trading software but rather
a development program that allows the trader to evaluate strategies in an
instant — literally. TradeSim cOvers the mechanical side of trading and,
providing your technique can be coded using MetaStock language or
BullScript, you are in business. (If you are not interested in learning how to
code, you are neglecting one of the most important elements within your
strategy blueprint — the development phase. Trading is only the end result of
development through repeated testing, whether this is done<p>NAB shares was</p><p>$31.98 and the RSI was 59.794 per cent. The process for determining the</p><p>level of neutrality is as follows:</p><p>1 subtract 50 from RSI value</p><p>RSI(Close,14)-50</p><p>59.794 - 50 = 9.794</p><p>2 divide by 100 to restore ratio</p><p>(RSI(Close,14)-50)/100</p><p>9.794 ÷ 100 = 0.09794</p><p>3 multiply ratio by closing price</p><p>Close*((RSI(Close,14)-50)/100)</p><p>31.98 × 0.09794 = 3.1321212</p><p>4 subtract total ratio value from closing price</p><p>Close-(Close*((RSI(Close,14)-50)/100))</p><p>31.98 - 3.1321212 = 28.8479</p><p>Dollar value of neutral point = $28.848</p><p>As BullCharts restricts monetary values to three decimal places, the RSI on</p><p>price rounds out to $28.848.</p><p>Although the following is a little premature in regard to the ongoing</p><p>discussion, while we are in mathematics mode I will also cOver determining</p><p>Oversold and Overbought regions.</p><p>Determining the Oversold level</p><p>The process for determining the Oversold level is as follows:</p><p>1 subtract 30 from RSI value</p><p>RSI(Close,14)-30</p><p>59.794 - 30 = 29.794</p><p>2 divide by 100 to restore ratio</p><p>(RSI(Close,14)-30)/100</p><p>29.794 ÷ 100 = 0.29794</p><p>3 multiply ratio by closing price</p><p>Close*((RSI(Close,14)-30)/100)</p><p>31.98 × 0.29794 = 9.5281212</p><p>4 subtract total ratio value from closing price</p><p>Close-(Close*((RSI(Close,14)-30)/100))</p><p>31.98 - 9.5281212 = 22.4518</p><p>Dollar value of Oversold level = $22.452</p><p>The Oversold region in figure 2.6 has a value of $22.45, as shown by the</p><p>calculation process above.</p><p>Determining the Overbought level</p><p>Please bear with me for one moment, as I am sure that you will pick up on</p><p>the point I am trying to highlight as we go along. Some people struggle once</p><p>negative values are included in the calculation process. So far it has simply</p><p>been a case of changing ‘50’, which defines the boundary between bullish</p><p>and bearish activity, to ‘30’ in order to define the Oversold region. This is a</p><p>relatively straightforward process. In order to identify the Overbought region,</p><p>you simply replace ‘30’ with ‘70’.</p><p>Using the process as discussed so far, you will regularly experience</p><p>negative relative strength values. To clarify what I am referring to, the RSI in</p><p>the example above has a value of 59.794 per cent. If you subtract 70, which</p><p>represents the Overbought region, from 59.794 per cent, you will produce a</p><p>negative result — minus 10.206 to be precise. This may lead you to believe</p><p>that you have made an error — after all, you are aiming to define the</p><p>Overbought region, which is well above the current price behaviour. Your</p><p>misconceptions may lead you to believe that you will place the Overbought</p><p>value below the Oversold regions, because you are subtracting a much larger</p><p>figure. This can all become a bit confusing to start with, so rather than trying</p><p>to explain how this works, I have again included the calculations so you can</p><p>see how the process works and feel comfortable with it. The process is as</p><p>follows:</p><p>1 subtract 70 from RSI value</p><p>RSI(Close,14)-70</p><p>59.794 - 70 = -10.206</p><p>2 divide by 100 to restore ratio</p><p>(RSI(Close,14)-70)/100</p><p>-10.206 ÷ 100 = -0.10206</p><p>3 multiply ratio by closing price</p><p>Close*((RSI(Close,14)-70)/100)</p><p>31.98 × -0.10206 = -3.263878</p><p>4 subtract total ratio value from closing price</p><p>Close-(Close*((RSI(Close, 14)-70)/100))</p><p>31.98 - -3.263878 = 35.2438</p><p>Dollar value for Overbought level = $35.244</p><p>The Overbought region has a value of $35.24. This is identical to the value</p><p>shown in figure 2.6 — as it must be. Once the relative strength index drops</p><p>below 50 per cent, negative values in both Overbought and Oversold</p><p>calculations will be encountered; however, this has no impact on the Overall</p><p>results. When you subtract a negative value from the closing price, which is</p><p>always greater than zero, this has the same effect as adding the total ratio to</p><p>the closing price, not subtracting. As the Overbought region will always</p><p>produce a greater negative value than the Oversold region, it will always</p><p>produce the higher dollar value of the two.</p><p>Figure 2.6 shows how these neutral, Oversold and Overbought levels have</p><p>been added to the NAB chart. The value of 50 per cent now assumes a</p><p>dynamic nature, while the original RSI value is replaced by price action. I</p><p>have also included the Overbought and Oversold regions directly onto price</p><p>action in figure 2.6; however, at this stage I have done this in order to</p><p>confirm the calculation process rather than for discussion purposes. I have</p><p>extended the number of periods normally applied for reasons of clarity only</p><p>at this stage.</p><p>Confirming the levels</p><p>Plotting the RSI value of 50 per cent onto price action is one thing, but the</p><p>accuracy of the concept also needs to be confirmed. If the concept is</p><p>performing as expected, a 50 per cent crossOver of the RSI should coincide</p><p>with a corresponding crossOver on price action. There is little point in</p><p>proceeding further until the basics are right. As figure 2.6 shows, the</p><p>crossOver points coincide. You now have the cornerstone required for</p><p>applying the relative strength index on price action in its entirety.</p><p>One point you must remember at this point is that with conventional</p><p>indicator application a 10 per cent change in price represents a 20 per cent</p><p>change in indicator value, as the indicator range has been effectively halved</p><p>by focusing on a value either side of neutral, or zero, rather than on the full</p><p>range of the index. This means you should not expect the percentage</p><p>movement experienced by price and by the indicator to be the same. This</p><p>occurs as you multiply the closing price and ratio as determined by the</p><p>percentage offset — that is, 50 per cent in the case of the neutral point — in</p><p>order to define a full monetary value.</p><p>I will not explore this relationship too deeply here, but I think you can get</p><p>the idea. As long as you can understand the basic relationship that exists</p><p>between the conventional relative strength index and its new price-based</p><p>cousin, that is all that matters at this point. It’s a bit like refilling your car at</p><p>the local service station — as long as you know whether you should use</p><p>unleaded or premium, that’s all that really matters. The specific octane level</p><p>is largely irrelevant to your everyday driving needs.</p><p>On figure 2.7, I have selected two points at random — one bullish and one</p><p>bearish. The RSI value of 50 per cent on price at the point to the left of the</p><p>chart has a value of $30.34 while the closing price is $26.20. Using the</p><p>BullCharts percentage tool, you can quickly ascertain that the bearish</p><p>dominance of price action is approximately 15.664 per cent. By rights, if the</p><p>above theory is correct, this should be half of the total amount that the</p><p>conventional RSI is below the value of 50 per cent. At this point, the RSI is</p><p>31.326 per cent below the 50 per cent region — which is 0.001 per cent off</p><p>being precisely double 15.664 per cent. Considering that most programs</p><p>round to partial values, it would be fair to accept that the conventional</p><p>relative strength index is exactly double the value of its price-related cousin.</p><p>Figure 2.7: confirming relative strength index and price relationship</p><p>It is only logical to also examine bullish activity for any anomalies in</p><p>application. At the second point shown on figure 2.7, the RSI value of 50 per</p><p>cent on price is 9.794 per cent below price action. The conventional RSI</p><p>should therefore be double this amount above the 50 per cent level — which,</p><p>at 19.588 per cent, it is exactly.</p><p>Creating the channels</p><p>Having successfully adapted the relative strength index to price action and</p><p>confirmed this adaptation, the time had come to check that my theories on</p><p>Overbought and Oversold levels were equally legitimate. While the coding</p><p>process later proved relatively straightforward, many methods of coding were</p><p>tested before a simple and accurate combination was finally identified. Rather</p><p>than repeat the entire calculation process, I have only included the final</p><p>formulas below. As you can see (and as already shown above), you simply</p><p>replace 50 with 70 for Overbought and 50 with 30 for Oversold.</p><p>Hence, the formula for Overbought is:</p><p>Close-(Close*((RSI(Close,14)-70)/100));</p><p>While the formula for Oversold is:</p><p>Close-(Close*((RSI(Close,14)-30)/100));</p><p>In order to confirm the validity of the Overbought and Oversold regions on</p><p>price action, it made sense to test the process on a more volatile stock, so</p><p>EquiGold (EQI) was selected in preference to the more sedate banking stocks</p><p>used for initial development.</p><p>Wilson Relative Price Channel</p><p>I would like to introduce the Wilson Relative Price Channel. The following</p><p>chart, shown in figure 2.8, reflects my first successful attempt at adapting the</p><p>relative strength index to price action. While it may not be Overly apparent,</p><p>the upward crossing of 70 per cent by the conventional RSI shown on figure</p><p>2.8 coincides with the upward crossOver of price action on the chart above,</p><p>which is now the upper cord of the newly created relative price channel. On</p><p>figure 2.8, the following areas have been highlighted:</p><p>Figure 2.8: Wilson Relative Price Channel showing conventional</p><p>Overbought and Oversold level on price</p><p>A Equigold has experienced a minor move above the classic Overbought</p><p>region of 70 per cent. This coincides with a corresponding move by price</p><p>action above the upper cord of the relative price channel. While extreme</p><p>trading activity is brief, the closing price finishes sufficiently above the</p><p>upper cord of the relative price channel, coinciding with an extreme RSI</p><p>value. This must occur as you are effectively looking at the identical</p><p>indicator — it is just that one is now relative to price action.</p><p>B This time around EQI experiences an extreme and sustained push above</p><p>the upper cord of the relative price channel, which is again mirrored by</p><p>the RSI moving into the Overbought region.</p><p>C The bears force price action downward in what is potentially an</p><p>unsustainable sell-off. While the possibility that it is unsustainable is not</p><p>readily apparent when the relative strength is displayed as an index, the</p><p>relationship between internal strength and price action is more clearly</p><p>evident when the two are combined on the one scale.</p><p>The Wilson Relative Price Channel with neutral</p><p>zone</p><p>For the first time, confirmation indicators can be directly combined with price</p><p>action; however, while this was a major step forward with my analysis,</p><p>something was still missing, believe it or not. This style of analysis is</p><p>primarily a visual interpretation of trend development, but I was still missing</p><p>one vital ingredient. The solution finally came to me early one morning while</p><p>my trusty little Jack Russell was busy pumping out the ‘zzzzs’. How could I</p><p>sleep with all that racket?! What I was missing was a region of neutrality.</p><p>Indicators will generally have a region of neutral activity — for example, 50</p><p>per cent for the RSI. While I had defined it mathematically, it had no visual</p><p>relevance that I could relate to on my chart. I needed to incorporate this into</p><p>my trading tool in order for it to be complete.</p><p>The last step in the development process was to incorporate a colour system</p><p>for bullish and bearish regions. For this, where possible, I use BullCharts as I</p><p>believe it has a superior method of visual representation — better than</p><p>MetaStock and other comparable programs I have encountered to date. This</p><p>also highlights one of the reasons seasoned traders will use more than one</p><p>trading program. The choice of program is about finding answers to specific</p><p>problems, and to date I am unaware of any program that includes all of the</p><p>answers I seek, including the expensive ones.</p><p>The relative channel properties window in BullCharts, shown in figure 2.9,</p><p>allows you to soften the indicator. For the purposes of writing this book I</p><p>deactivated the kind of line displays shown in figures 2.6 and 2.8. I also</p><p>applied two shades of grey for printing purposes at the request of my editor;</p><p>however, I have since found that contrasting shades of grey, as opposed to the</p><p>more prominent colours available in the properties pane, work well with real-</p><p>time analysis due to the more subtle effect. While the channels are clearly</p><p>evident on the screen, they are not the dominant element, meaning they</p><p>actually complement price action rather than Overpower it. It may pay to</p><p>spend a little time finding a suitable balance in contrasts as its visual impact</p><p>can be varied and quite notable. The quality of your monitor and the selected</p><p>resolution are also worth some consideration as they will also influence the</p><p>final image.</p><p>Figure 2.9: relative channel properties window</p><p>Figure 2.10 is the end result of the transition process that has seen the relative</p><p>strength index evolve from a range-bound index at the bottom of the screen to</p><p>a dynamic strength channel relative to price action.</p><p>Figure 2.10: Wilson Relative Price Channel and relative strength index</p><p>Well, there you have it, the end result as I apply it today — well, one of them</p><p>anyway. The Wilson Relative Price Channel shown in figure 2.10 consists of</p><p>the following elements:</p><p>bullish band — this region defines the range of bullish activity between</p><p>55 per cent and 70 per cent</p><p>bearish band — this region defines the range of bearish activity between</p><p>30 per cent and 45 per cent</p><p>neutral zone — this region defines neutral activity between 45 per cent</p><p>and 55 per cent</p><p>cords — each band and the neutral zone incorporate upper and lower</p><p>cords that define the extremities of each region</p><p>Overbought activity — price action that occurs above the bullish band</p><p>Oversold activity — price action that occurs below the bearish band.</p><p>For continuity I display the conventional indicator at the bottom of the screen</p><p>in the identical manner as that applied to price action. This has allowed a very</p><p>visual approach to the analysis process where applying both indicators using</p><p>the same visual configuration allows you to compare both analysis tools</p><p>simultaneously, knowing that you are viewing the one set of parameters from</p><p>two perspectives.</p><p>Theoretical ways to trade price action using</p><p>channels</p><p>With conventional application, the RSI worked independently of price action,</p><p>requiring traders to define the relationship visually. This is ultimately reliant</p><p>upon experience and a solid understanding of the indicator’s application and</p><p>calculation process. The process is now somewhat different. As the RSI</p><p>defines internal strength or weakness driving trend development, by adapting</p><p>the same calculation process to price action and then making it relative to</p><p>general trend development, in theory you should be able to trade price action</p><p>as it moves between the bullish and bearish regions.</p><p>The following is not cast in stone and should be applied as a guide only, as</p><p>the method of application will depend on the underlying trend and strategy in</p><p>place. Some of you will be suggesting that the mere idea of trading a relative</p><p>price channel as a stand-alone technique — that is, without using any other</p><p>tools to confirm the entry or exit signal — is sheer lunacy. This is perfectly</p><p>acceptable and while you may decide not to trade a relative price channel as a</p><p>stand-alone technique, have you ever considered the significance of using a</p><p>confirmation tool that is profitable in its own right? Probably not; however,</p><p>you will never know the true effectiveness of your confirmation tool if you</p><p>do not explore this avenue. Put another way, why would you rely on a</p><p>confirmation tool that tests as unprofitable? Testing whether the relative price</p><p>channel can be successfully traded on its own does not imply that it must be</p><p>applied in this manner; however, what it could mean is that the confirmation</p><p>tool on its own is a profitable tool, which in turn can add confidence to your</p><p>approach and weight to your analysis.</p><p>Figure 2.11 shows the way I divide the channels into five separate</p><p>elements.</p><p>Figure 2.11: Wilson Relative Price Channel — the basics</p><p>These elements are as follows:</p><p>Sell long — you can often look to lock in profits when price action</p><p>moves beyond the upper cord of the bullish band, which is effectively an</p><p>RSI value greater than 70. The following conditions apply:</p><p>If price action has experienced a blow-off</p><p>— in other words, there</p><p>has been a significant jump or spike in price, often Over just one or</p><p>two days — you can look to lock in profits.</p><p>If price action has moved beyond the bullish band on parabolic</p><p>activity — that is, price action has accelerated exponentially —</p><p>again, you can look to lock in profits at the first signs of a reversal.</p><p>If price action has moved beyond the outer limits without</p><p>experiencing extreme trading activity, you can look to hold.</p><p>However, in this situation, you should tighten your trailing stops if</p><p>you think that more than a temporary pull back in price or</p><p>consolidation may be in progress.</p><p>Buy long/hold long — you should trade the stock from the long side</p><p>only when price action enters the bullish region. You can continue to</p><p>hold the stock for as long as price action continues to trade in this</p><p>particular zone, unless your specialised trailing stop closes the position.</p><p>In the unlikely event that you have not been stopped out, you would not</p><p>continue to hold the stock if price action drifted into the ‘sell short/hold</p><p>short’ region.</p><p>Wait for break-out — the neutral zone generally defines non-trending or</p><p>unbiased price action, so you should not consider opening a position</p><p>either long or short until price action moves beyond this region.</p><p>Obviously, trend reversals must go through the neutral zone on their</p><p>path to higher or lower values; however, you should be careful when</p><p>trading price action that, after tagging the outer extremities of the</p><p>relative price channel, is moving from bearish to bullish values and vice</p><p>versa. If price action has experienced a bullish tag of the upper cord of</p><p>the bullish band and reversed, it is relatively common for price action to</p><p>tag the upper cord of the bearish band and then pause and rebound, or</p><p>begin to drift. Likewise, if price action has experienced a bearish tag of</p><p>the lower cord of the bearish band and reversed, it is relatively common</p><p>for price action to tag the lower cord of the bullish band and then pause</p><p>and rebound in a bearish manner, or drift in the neutral region.</p><p>Sell short/hold long — the stock should be traded from the short side</p><p>only when price action first enters the bearish band. You should</p><p>continue to hold the stock for as long as price action continues to trade</p><p>in this particular band, unless your specialised short side trailing stop</p><p>closes the position. In the unlikely event that you have not been stopped</p><p>out, you should not continue to hold the stock if price action drifted into</p><p>the ‘buy long/hold long’ region.</p><p>Buy short — you can look to lock in profits by closing any open short</p><p>positions when price action moves beyond the lower cord of the bearish</p><p>band, which is equivalent to an RSI value of less than 30. Again, you</p><p>should keep the following in mind when price action extends beyond the</p><p>lower band of the channel:</p><p>If price action has experienced a pile-driver — in other words,</p><p>there has been a significant and sudden drop in price, often Over</p><p>just one or two days — you can look to lock in profits.</p><p>If price action has moved beyond the lower cord of the bearish</p><p>band on parabolic activity — that is, price action has accelerated</p><p>exponentially downward — again, you can look to lock in profits at</p><p>the first signs of a reversal.</p><p>If price action has moved beyond the lower cord of the bearish</p><p>band without experiencing extreme trading activity, you can look to</p><p>hold. However, in this situation, you should tighten your trailing</p><p>stops if you think that more than a temporary pull back in price or</p><p>consolidation may be in progress. Be aware that downtrends will</p><p>consolidate in a slightly different manner and often less frequently</p><p>than uptrends.</p><p>Before these regions can be used effectively in trading, whether the channels</p><p>are truly relative to price action needs to be confirmed. The following</p><p>benchmark, and the benchmarks in the following four chapters, will test the</p><p>effectiveness of the 14-period channel as a stand-alone tool.</p><p>Establishing a benchmark — relative price</p><p>channel</p><p>If you are going to establish whether any improvement can be identified after</p><p>adapting your indicators to price action, you need to have a benchmark that</p><p>preferably reflects accepted practice. Once this benchmark has been</p><p>established, you can then test whether it is possible to improve a confirmation</p><p>tool by adjusting the indicators to better suit a relative relationship — for</p><p>example, by adjusting the periods rather than just applying the default half-</p><p>lunar cycle value of 14 periods. Only then will you know for sure that not</p><p>only do your changes appear more effective on the surface, but the net</p><p>improvement also reflects a more consistent relationship. Trading results</p><p>should reflect positive change on a number of fronts if the concepts used are</p><p>truly relative. By the way, I never did work out what lunar cycles have to do</p><p>with trading.</p><p>Testing parameters for the benchmark</p><p>The testing parameters used to establish the benchmark against which any</p><p>further concepts that incorporate the price channels can be compared are as</p><p>follows:</p><p>channel periods — 14</p><p>entry trigger — close above the lower cord of the bullish band</p><p>exit trigger — close below the lower cord of the bullish band or a close</p><p>below the upper cord of the bullish band for price action that has moved</p><p>into the Overbought region.</p><p>The purpose of creating a benchmark is not to identify optimum performance</p><p>at this early stage; it is about the identification of a suitable point for further</p><p>comparison. The only way you can know whether you are improving or</p><p>travelling backwards is to have a suitable line drawn in the sand.</p><p>I decided to test accepted practice by first using a 14-period relative price</p><p>channel, which is the price equivalent of a 14-period RSI. Not only would</p><p>this reflect the approach applied by the majority of traders who use the RSI, it</p><p>would also create a suitable benchmark for ongoing development. It is fine</p><p>for me to waffle on about improvements; however, the only method I know</p><p>of to resolve any doubts Over effectiveness is to open the approach up to</p><p>scrutiny and, other than real-time application, there is no better way to</p><p>achieve this than to place the strategy at the mercy of a program such as</p><p>TradeSim. The bulldust stops when TradeSim starts — it shows no bias, only</p><p>results produced by extensive and repeated application. The result is not</p><p>gilded, nor tailored toward protecting delicate egos. If the strategy is pure</p><p>garbage, the testing results will reflect the rubbish that has been entered.</p><p>Another criticism may be that all testing has been performed on historical</p><p>data; however, this is all traders ever have to work with when wanting to</p><p>evaluate techniques on vast amounts of data. While real-time application may</p><p>be the only true test, it takes many years and hundreds of trades to amass</p><p>sufficient data in order to draw a rational conclusion. Even after many years</p><p>of real-time application, depending on the research capabilities, these results</p><p>could still be vastly inaccurate. I have been developing and applying these</p><p>techniques Over an extended period; however, it is doubtful that my level of</p><p>understanding and advancement would be where it is today without a decent</p><p>research program. Real-time application is no guarantee that you will get</p><p>things right, and it can take ten years to find out you were wrong.</p><p>Testing results for the benchmark</p><p>The test results for the benchmark for the relative price channel, also shown</p><p>in table 2.2, were as follows:</p><p>Table 2.2: Wilson Relative Price Channel — benchmark</p><p>Profit summary</p><p>Profit status PROFITABLE</p><p>Starting capital $25 000.00</p><p>Finishing capital $78 091.47</p><p>Maximum equity/(date) $53 091.47 (7/10/2005)</p><p>Minimum equity/(date) −$5428.42 (13/12/2001)</p><p>Gross trade profit $87 878.26 (351.51%)</p><p>Gross trade loss −$34786.78 (−139.15%)</p><p>Total net profit $53 091.47 (212.37%)</p><p>Average profit per trade $270.87</p><p>Profit factor 2.5262</p><p>Profit index 60.41%</p><p>Total transaction cost $7840.00</p><p>Total slippage $0.00</p><p>Daily compound interest rate 0.0543%</p><p>Annualised compound interest</p><p>rate 21.8926%</p><p>Trades processed 3701</p><p>Trades taken 196</p><p>Partial trades taken 0</p><p>Trades rejected 3505</p><p>Winning trades 47 (23.98%)</p><p>Losing trades 149 (76.02%)</p><p>Largest winning trade/(date) $13 489.06 (17/12/2004)</p><p>Largest losing trade/(date) −$1421.38 (31/05/2002)</p><p>Average winning trade $1869.75</p><p>Average losing trade −$233.47</p><p>Average win/average loss 8.0086</p><p>Trade duration statistics (all trades)</p><p>Maximum trade duration 581 (days)</p><p>Minimum trade duration 6 (days)</p><p>Average trade duration 57 (days)</p><p>Winning trades</p><p>Maximum trade duration 581 (days)</p><p>Minimum trade duration 7 (days)</p><p>Average trade duration 175 (days)</p><p>Losing trades</p><p>Maximum trade duration 126 (days)</p><p>Minimum trade duration 6 (days)</p><p>Average trade duration 19 (days)</p><p>Consecutive trade statistics</p><p>Maximum consecutive winning trades 5</p><p>Maximum consecutive losing trades 16</p><p>Average consecutive winning trades 1.57</p><p>Average consecutive losing trades 4.97</p><p>Relative drawdown</p><p>Maximum dollar drawdown/(date) $3160.28 (11/10/2002)</p><p>Maximum percentage drawdown/(date) 12.3100% (10/08/2001)</p><p>Absolute (peak to valley) dollar drawdown</p><p>Maximum dollar drawdown $7711.89 (22.2700%)</p><p>Capital peak/(date) $34 635.79 (17/05/2002)</p><p>Capital valley/(date) $26 923.90 (14/03/2003)</p><p>The concept in its raw form produced too many signals. Over the</p><p>testing period we experienced a total of 3701 potential trades, which is far</p><p>too many for real-time application. This needs to be reduced.</p><p>The concept produced 16 consecutive unprofitable trades, which is well</p><p>below the tolerance levels of most traders.</p><p>The percentage of profitable trades was very low with a dismal 23.98</p><p>per cent success rate achieved. This is unacceptably low and requires a</p><p>significant improvement before it would be practical for real-time</p><p>application. As discussed in The Next Step to Share Trading Success, the</p><p>ideal success rate is the magic mark of 50 per cent. The closer you push</p><p>the success rate toward the 50 per cent region, the higher the probability</p><p>of real-time survival.</p><p>The peak to valley drawdown is unacceptable at 22.27 per cent.</p><p>The annualised percentage return is an acceptable 21.89 per cent, so the</p><p>concept is theoretically profitable. I say ‘theoretically’ profitable at this</p><p>stage simply because the success rate is so low. I know of very few</p><p>traders who could apply this particular concept in its current form Over</p><p>the longer term. While the continued run of losses is fine in theory, very</p><p>few of us have the ability to stay with a strategy that repeatedly produces</p><p>negative results. You will notice that I am not using the word ‘strategy’ at</p><p>this point with regard to the benchmark. This is simply because it is not</p><p>yet a strategy, just a concept that may have potential at some point down</p><p>the track.</p><p>The average loss was just $233.47, which is more than acceptable,</p><p>while the largest losing trade experienced during the testing period was</p><p>definitely passable at −$1421.38.</p><p>Net returns versus equity curve and yearly profits</p><p>As discussed in chapter 1, one of the biggest concerns I regularly have when</p><p>I’m asked to review a private strategy for someone is the person’s primary</p><p>focus on the net return — and we all can be guilty of this at some point.</p><p>However net profit can only be achieved if the strategy can be traded</p><p>successfully in real time. In this case, the concept has been comprehensively</p><p>tested on the top 300 stocks since January 2000, and it has produced an</p><p>acceptable net return. Some people may simply look at the dollar value and</p><p>say, ‘Look, I could turn $25 000 into $78 091.47’. This is an incorrect and a</p><p>somewhat irresponsible conclusion to draw. The concept only produces</p><p>legitimate entry signals 23.98 per cent of the time, which in reality means that</p><p>76.02 per cent — or three-quarters of all trades — will cost trading capital.</p><p>You could also expect a string of losses while having to identify legitimate</p><p>signals from a vast number of dodgy triggers. While the concept may be</p><p>theoretically profitable, at this point of development it is impractical to apply</p><p>in real time.</p><p>A look at the closed equity curve, shown in figure 2.12, reveals a number of</p><p>undesirable features of the concept at this early stage. The bulk of profits</p><p>occurred in the space of a relative short period in 2004. The remaining four</p><p>years produced consistent results with the exception of 2000 and 2001, with</p><p>losses of 14.79 per cent and 8.11 per cent respectively. At best you could</p><p>conclude that a 14-period relative price channel is a somewhat hit-and-miss</p><p>method of analysis. It’s as though the RSI only functions under certain</p><p>conditions and, beyond this, extended periods of incompatibility can be</p><p>expected. This incompatibility no doubt contributes to the low success rate of</p><p>the signals regularly identified.</p><p>Figure 2.12: closed equity curve for 14-period relative price channel</p><p>A closer look at the yearly profit graph, shown in figure 2.13, confirms that</p><p>the bulk of the profit achieved is limited to 2002 and 2004, as just mentioned;</p><p>however, a degree of consistency does exist with exception of 2000 and</p><p>2001. The question remains as to whether a trader could tolerate a portfolio</p><p>being in the red for almost two and a half years. The amount of trust that a</p><p>trader would have in a 14-period index by the time a positive outcome was</p><p>achieved would in most cases be non-existent.</p><p>Figure 2.13: yearly return for 14-period relative price channel</p><p>The message from this is never let the dollar return dictate the type of</p><p>strategy implemented. While this is only a benchmark for ongoing</p><p>development, it should never be applied in real time.</p><p>Conclusions to draw at this early stage</p><p>Based on the testing results from the relative price channel benchmark, the</p><p>default value of 14 periods is far from relative to trend development and</p><p>general market behaviour Over the long term. This is borne out by the limited</p><p>surge in efficiency Over a relatively short period in late 2004, especially</p><p>when the fact that the market trended strongly since March 2003 and its</p><p>exceptional success rate of 23.98 per cent is considered. If I said to you,</p><p>‘Here is the greatest confirmation tool in existence — oh, by the way, it’s</p><p>reliable one time in four’, you would tell me to take a hike. If I insert the</p><p>same indicator into a software program with no instructions on interpretation,</p><p>it suddenly gains new-found respect. Why? Because at face value there looks</p><p>to be a consistent correlation between the indicator and trend development.</p><p>While retrospective analysis may suggest a high level of correlation between</p><p>the indicator and price action, real-time application can be far more</p><p>subjective once you move beyond the blatantly obvious points in</p><p>development. However, the adaptation of the indicator to price action and the</p><p>creation of a benchmark now opens up the potential to mould previously</p><p>ineffective techniques to actual trend development.</p><p>Chapter 3 looks at adapting an open oscillator to price action.</p><p>Coding for chapter 2</p><p>Wilson Relative Channel Index — BullCharts</p><p>[Target=Percent; author=Wilson, Leon]</p><p>Periods:=input("RSI Periods", 34, 1, 250);</p><p>Smoothing:=input("Smoothing", 1, 1, 55);</p><p>Trigger:=input("Trigger Line", 1, 1, 55);</p><p>Value2:=input("Over Bought", 70, 50, 99);</p><p>Value3:=input("Over Sold", 30, 1, 50);</p><p>Value4:=input("Upper Neutral Zone", 55, 50, 99);</p><p>Value5:=input("Lower Neutral Zone", 45, 1, 50);</p><p>Index:= Ma(RSI(CLOSE,Periods),Smoothing,E);</p><p>Trigg:=Ma(Ma(RSI(CLOSE,Periods),Smoothing,E),Trigger,E);</p><p>[Color=Black;Linestyle=Solid]</p><p>Index;</p><p>[Color=Blue;Linestyle=dash]</p><p>Trigg;</p><p>[Color= Light Slate Gray; Linestyle=Dotted]</p><p>value3; value2;</p><p>[name=1</p><p>st</p><p>Zone1]</p><p>Value5;</p><p>[name=2</p><p>nd</p><p>Zone2]</p><p>value4;</p><p>[name=Bullish Zone; linestyle=Fill;color=Light Slate Gray ]</p><p>Value2; Value4;</p><p>[name=Neutral Zone; linestyle=Fill;color=White]</p><p>value5; Value4;</p><p>[name=Bullish Zone; linestyle=Fill;color=Dark Grey]</p><p>Value3; Value5;</p><p>{Copyright Leon Wilson, 2005}</p><p>Wilson Relative Price Channel — BullCharts</p><p>[Target=Price; author=Wilson, Leon]</p><p>Periods:=input("Channel Periods",34,1,250);</p><p>Smoothing:=input("Smoothing",</p><p>1, 1, 55);</p><p>Value2:=input("Over Bought", 70, 50, 99);</p><p>Value3:=input("Over Sold", 30, 1, 50);</p><p>Value4:=input("Upper Neutral Zone", 55, 50, 99);</p><p>Value5:=input("Lower Neutral Zone", 45, 1, 50);</p><p>[Color= Light Slate Gray; Linestyle=Dotted]</p><p>OB:=Ma((RSI(Close,Periods)-Value2),Smoothing,E);</p><p>OS:=Ma((RSI(Close,Periods)-Value3),Smoothing,E);</p><p>NZU:=Ma((RSI(Close,Periods)-Value4),Smoothing,E);</p><p>NZL:=Ma((RSI(Close,Periods)-Value5),Smoothing,E);</p><p>{Show Results}</p><p>[name=OverSold;]</p><p>Close-(Close*(OS/100));</p><p>[name=OverBought;]</p><p>Close-(Close*(OB/100));</p><p>[name=NeutUp;]</p><p>Close-(Close*(NZU/100));</p><p>[name=NeutLower;]</p><p>Close-(Close*(NZL/100));</p><p>{Upper Channel}</p><p>[name=Upper; linestyle=Fill;color=Light Slate Gray]</p><p>Close-(Close*(NZU/100)); Close-(Close*(OB/100));</p><p>{Neutral Zone}</p><p>[name=Zone; linestyle=Fill;color=White]</p><p>Close-(Close*(NZU/100)); Close-(Close*(NZL/100));</p><p>{Lower Channel}</p><p>[name=Lower ; linestyle=Fill;color=Dark Grey]</p><p>Close-(Close*(NZL/100)); Close-(Close*(OS/100));</p><p>{Copyright Leon Wilson, 2005}</p><p>Wilson Relative Channel Index — MetaStock</p><p>Periods:=Input("Periods", 1, 250, 34);</p><p>Smoothing:=Input("Smoothing", 1, 55, 1);</p><p>Value2:=Input("Over Bought", 50, 99, 70);</p><p>Value3:=Input("Over Sold", 1, 50, 30);</p><p>Value4:=Input("Upper Neutral Zone", 50, 99, 55);</p><p>Value5:=Input("Lower Neutral Zone", 1, 50, 45);</p><p>Index:= Mov(RSI(CLOSE,Periods),Smoothing,E);</p><p>Trigg:=Mov(Mov(RSI(CLOSE,Periods),Smoothing,E),1,E);</p><p>Index; Trigg;</p><p>Value3; Value2; Value5; Value4;</p><p>{Copyright Leon Wilson, 2005}</p><p>Please note: an adjustable trigger line is not possible in MetaStock with some</p><p>formulas due to 6 fields of maximum input</p><p>Wilson Relative Price Channel — MetaStock</p><p>Periods:=Input("Channel Periods", 1,250,34);</p><p>Smoothing:=Input("Smoothing", 1, 55, 1);</p><p>Value2:=Input("Over Bought", 50, 99, 70);</p><p>Value3:=Input("Over Sold", 1, 50, 30);</p><p>Value4:=Input("Upper Neutral Zone", 50, 99, 55);</p><p>Value5:=Input("Lower Neutral Zone", 1, 50, 45);</p><p>OB:=Mov((RSI(CLOSE,Periods)-Value2),Smoothing,E);</p><p>OS:=Mov((RSI(CLOSE,Periods)-Value3),Smoothing,E);</p><p>NZU:=Mov((RSI(CLOSE,Periods)-Value4),Smoothing,E);</p><p>NZL:=Mov((RSI(CLOSE,Periods)-Value5),Smoothing,E);</p><p>CLOSE-(CLOSE*(OS/100));</p><p>CLOSE-(CLOSE*(OB/100));</p><p>CLOSE-(CLOSE*(NZU/100));</p><p>CLOSE-(CLOSE*(NZL/100));</p><p>{Copyright Leon Wilson, 2005}</p><p>Relative Price Channel — TradeSim, MetaStock</p><p>Pds:=14;</p><p>Index:=RSI(Close,Pds);</p><p>NZU:=Index -55;</p><p>NZL:=Index -45;</p><p>UC:=Index -70;</p><p>NeutUp:=CLOSE-(CLOSE*(NZU/100));</p><p>OB:=CLOSE-(CLOSE*(UC/100));</p><p>NeutL:= CLOSE-(CLOSE*(NZL/100));</p><p>EntryTrigger := CLOSE>NeutUp AND Ref(CLOSE,-1)<=Ref(NeutUp,-1);</p><p>EntryPrice := OPEN;</p><p>ExitTrigger := (CLOSE<NeutUp AND (Ref(CLOSE,-1) >=Ref(NeutUp,-1))) OR</p><p>(CLOSE<OB AND (Ref(CLOSE,-1) >=Ref(OB,-1)));</p><p>ExitPrice := OPEN;</p><p>InitialStop:=0;</p><p>{Copyright Leon Wilson, 2005}</p><p>Chapter 3</p><p>The emergence of relative adaptive analysis</p><p>So far I have discussed what would probably be generally accepted as the</p><p>most popular indicator in use today; however, this does not imply that the</p><p>first option is necessarily the most appropriate. Indeed, I believe the relative</p><p>strength index is far from ideal. My dissatisfaction with the RSI stems from</p><p>the result being an averaged value, whereas on occasions I wanted to see raw</p><p>behaviour. If I wanted to smooth the indicator, I could incorporate this</p><p>function at the end of the calculation process.</p><p>Indicators, explorations and strategies usually stem from a range of</p><p>questions traders ask in regard to price and trend development. The solutions</p><p>are seldom singular and will vary greatly — depending, for example, on the</p><p>trader’s personal experience, access to quality technical trading programs and</p><p>the ability to derive a suitable answer from such software. A seasoned trader</p><p>using TradeStation will most likely arrive at a different and probably more</p><p>complex conclusion to a problem than an inexperienced trader who may use</p><p>BullCharts or MetaStock in the early stages of his or her trading career.</p><p>While the same solution may be possible with all three programs, the</p><p>seasoned trader with years of practical trading experience may quickly</p><p>identify a practical solution not readily apparent to those less experienced</p><p>traders surrounding them.</p><p>Tip</p><p>Before you can develop an indicator further, you must first understand both its strengths</p><p>and its weaknesses. Strengths and weaknesses are not only defined by the calculation</p><p>process but also by time and the manner in which the information is presented. An error in</p><p>any one of these areas leads to an inaccurate indicator that is of little benefit in real time.</p><p>My problem with the unwanted effects of smoothing initially led to the</p><p>creation of the internal strength oscillator. While the indicators appear similar</p><p>visually, the scaling differed due to my application of an oscillator style</p><p>calculation.</p><p>Chapter 2 looked at adapting a range-bound index to price action. Using the</p><p>example of the internal strength oscillator, in this chapter I will work through</p><p>the process of converting an oscillator into a range-bound indicator, thereby</p><p>creating the internal strength index. The conversion of the oscillator into an</p><p>index ultimately meant the transformation process could be completed, as the</p><p>index could then be applied as a price channel. I will work through the</p><p>transition process Over the course of this chapter.</p><p>Converting an open oscillator into a range-</p><p>bound index</p><p>The adaptation to price action process discussed in chapter 2 can only be</p><p>applied to range-bound oscillators or indices where the extreme values are</p><p>capped; however, a true oscillator is not capped. True oscillators are</p><p>relatively common and, as the name suggests, oscillate around a central axis.</p><p>Hence, in order to adapt an open oscillator to price action, the oscillator first</p><p>needs to be converted into a range-bound index. The range of potential</p><p>movement in the oscillator needs to be limited or capped, otherwise in</p><p>extreme conditions the price channel will push well beyond the scope of</p><p>current trading activity, and so compress surrounding price action. Once the</p><p>possible range of the original oscillator has been capped, the same process</p><p>discussed in chapter 2 can be applied.</p><p>Internal strength oscillator to internal strength</p><p>index</p><p>The coding used to create the internal strength oscillator is below:</p><p>Up:=Sum(If(CLOSE>Ref(CLOSE,-1),CLOSE-Ref(CLOSE,-1),0),21);</p><p>Down:=Sum(If(Ref(CLOSE,-1)>CLOSE,Ref(CLOSE,-1)-CLOSE,0),21);</p><p>((Up - Down)/(Up + Down)) * 100</p><p>The above coding is a relatively common approach when it comes to creating</p><p>an oscillator. While this coding works in the manner intended, the problem</p><p>with adapting the above formula is that a 0 to 100 index needs to be</p><p>developed from a segmented calculation process — that is, ‘(Up –</p><p>Down)/(Up + Down)’. This means some lateral thinking is required. Range-</p><p>bound indices usually look to compare single values, typically one bullish</p><p>against one bearish. The independent subtraction and addition of additional</p><p>values occurs prior to the percentage calculation step, meaning the incorrect</p><p>information is presented for the purposes of creating an index. In order for</p><p>you to correctly turn the above oscillator into an index, you also need to</p><p>understand where you can go wrong.</p><p>Below is the coding for the incorrect conversion of an oscillator into a</p><p>range-bound index:</p><p>Up:=Sum(If(CLOSE>Ref(CLOSE,-1),CLOSE-Ref(CLOSE,-1),0),21);</p><p>Down:=Sum(If(Ref(CLOSE,-1)>CLOSE,Ref(CLOSE,-1)-CLOSE,0),21);</p><p>100-(100/(1+((Up - Down)/(Up + Down))))</p><p>The natural thought process may lead some to consider that the coding above</p><p>is the correct process for index construction. However, while it may appear</p><p>technically correct on the surface, the index will never plot correctly while</p><p>the final line remains fragmented. I transformed the above formula using the</p><p>underlying principles of index construction to develop the internal strength</p><p>index.</p><p>I revised the final line of the calculation process above as follows:</p><p>'(Up - Down)/(Up + Down)' to '(Up/Down)'</p><p>Having removed the fragmented calculation process, a range-bound index</p><p>can now be</p><p>created. The coding can also be simplified as follows:</p><p>Up:= sum(max(CLOSE - Ref(CLOSE,-1),0),21);</p><p>Down:= sum(max(Ref(CL0SE,-1) - CLOSE,0),21);</p><p>100-(100/(1+(Up/Down)));</p><p>While I have not jeopardised the integrity of the original indicator, I have</p><p>effectively redefined the calculation process in order to produce a range-</p><p>bound index, and so created the internal strength index. As the final</p><p>calculation process differs from the original and it is presented in an index</p><p>rather than an oscillator format, I took the liberty of naming it accordingly.</p><p>Having successfully limited the range of an oscillator by creating the internal</p><p>strength index, the next task is to identify comparable Overbought and</p><p>Oversold regions in order to more accurately reflect everyday application.</p><p>When converting an oscillator to an index, you must not infringe the integrity</p><p>of the original calculation process.</p><p>Comparing Overbought and Oversold regions</p><p>As can be seen in figure 3.1, the newly created index is consistent with the</p><p>original oscillator. All peaks and troughs coincide, while general index</p><p>development mirrors the original. The scaling of the new indicator is 0 per</p><p>cent to 100 per cent, with a point of neutrality at 50 per cent that coincides</p><p>with the reference of the original. Further, Overbought and Oversold regions</p><p>correspond with the original oscillator as both indicators tag the peaks</p><p>simultaneously. This should be the case because, ideally, the original</p><p>calculation process has not been altered, other than the introduction of an</p><p>index-based process for the same mathematical calculation.</p><p>Figure 3.1: internal strength oscillator and internal strength index</p><p>comparison</p><p>To reiterate, before you attempt to adapt any indicator to price action, you</p><p>need to be sure of two points:</p><p>The indicator you are going to apply is range bound — that is, the values</p><p>are capped.</p><p>Any oscillator that has been adapted to an index format is consistent</p><p>with the original.</p><p>Internal strength index</p><p>As well as being able to convert the original oscillator into an index, it is</p><p>important also to understand how the indicator works and what it represents.</p><p>Fully understanding how an indicator works and what it represents will help</p><p>you to fully comprehend much of what will be discussed in the following</p><p>chapters. It really is surprising just how many people will use an indicator</p><p>such as the relative strength index yet have no idea of what it represents.</p><p>People will occasionally say to me, ‘I don’t know how you think of these</p><p>things’. I think of them because I take the time to fully understand what a</p><p>concept represents and how this will affect my trading. Only when I can</p><p>grasp the mechanics of an indicator can I successfully develop it further —</p><p>otherwise, any development is primarily reliant on guesswork, which is</p><p>seldom compatible with capital retention. In reality, anyone is capable of</p><p>developing new strategies, if they only took the time to think about what is in</p><p>front of them and how it relates to the questions they have and the answers</p><p>they seek.</p><p>In looking at the internal strength index, I will briefly go Over some ground</p><p>previously discussed in The Business of Share Trading in order to refresh the</p><p>grey matter, although I have slightly adapted the material to incorporate it</p><p>into this discussion.</p><p>The internal strength index (ISI) appears and reacts in the same manner as</p><p>most other indicators. My reason for developing the internal strength index</p><p>was not with the intention of designing a superior indicator but rather to</p><p>control the averaging process. The ISI is so named as it evaluates the bias</p><p>between bullish and bearish activity without the smoothing effect of the RSI.</p><p>I will attempt to keep the explanation of the internal strength index simple</p><p>without detracting from your potential understanding of the inner workings.</p><p>The ISI calculates the change in price action between bullish and bearish</p><p>activity separately Over a specified number of periods usually determined by</p><p>the user. By separately, I mean it sums the range experienced by all up days</p><p>(those with higher closes), while also summing the range encountered by all</p><p>down days (those days with lower closing prices) independently. It then</p><p>divides the total degree of bullish movement by the degree of bearish activity</p><p>in order to define market bias as a percentage.</p><p>The ISI lends itself well to trendline placement and is prone to experiencing</p><p>divergences.</p><p>The internal strength index and indicator values</p><p>The ISI is normally displayed as an index and is range bound between the</p><p>values of 0 and 100 per cent. As with the RSI, there are two arbitrary values</p><p>of 30 per cent and 70 per cent for the ISI, where values less than 30 per cent</p><p>are considered Oversold and values greater than 70 per cent are considered</p><p>Overbought. However, as the indicator is slightly more responsive than the</p><p>RSI, I commonly apply values of 25 per cent and 75 per cent for these</p><p>regions. I prefer to consider these regions as areas of Overextension with</p><p>regard to price action. ISI values also indicate the following:</p><p>A value of 50 per cent indicates that the range experienced by the</p><p>closing price on up days (higher closes) is comparable to the range</p><p>experienced by the closing price on down days (lower closes) Over the</p><p>same period.</p><p>A value of 70 per cent reflects a 40 per cent bias in favour of the range</p><p>experienced by the closing price on up days (higher closes) compared to</p><p>the range experienced by the closing price on down days (lower closes)</p><p>Over the same time period.</p><p>A value of 30 per cent reflects a 40 per cent bias in favour of the range</p><p>experienced by the closing price on down days (lower closes) compared</p><p>to the range experienced by the closing price on up days (higher closes)</p><p>Over the same time period.</p><p>This is shown in table 3.1.</p><p>Table 3.1: internal strength index values and percentage bias</p><p>Indicator value % bias</p><p>0 100.0% (bearish)</p><p>10 80.0%</p><p>20 60.0%</p><p>30 40.0%</p><p>40 20.0%</p><p>50 0.0% (neutral)</p><p>60 20.0%</p><p>70 40.0%</p><p>80 60.0%</p><p>90 80.0%</p><p>100 100.0% (bullish)</p><p>Indicator values and trend conclusion</p><p>Obviously, the ISI will gravitate around the 50 per cent value when price</p><p>action is not trending strongly. What will happen then during an uptrend?</p><p>Naturally, the ISI will be concentrated in the top half to top two-thirds of the</p><p>index value. All healthy trends generally experience regular pauses in price</p><p>action, and during these pauses the ISI will return to the neutral region of 50</p><p>per cent, crossing the Overbought value of 70 per cent — potentially closing</p><p>people out of their trades. Every time price action slows or retraces, the ISI,</p><p>by design, must gravitate back to the 50 per cent region. The opposite is true</p><p>for downtrends. As the bulk of ISI activity will be within the bottom half to</p><p>bottom two-thirds of index value, this will naturally produce negative values.</p><p>Such retracement behaviour by the ISI does not guarantee trend conclusion.</p><p>This is where some care needs to be taken when assessing divergences, as</p><p>the ISI will gravitate back to a more neutral value as prices consolidate.</p><p>Usually, I will only act on divergences when prices are still trending and will</p><p>treat divergences that occur during periods of consolidation with caution.</p><p>The following guidelines can be applied to the ISI:</p><p>An upward crossOver from the Oversold region can be considered a buy</p><p>signal. (Ensure that the ISI is accompanied by price action to minimise</p><p>the possibility that the ISI is just gravitating back to a more neutral</p><p>value.)</p><p>A downward crossOver from the Overbought region can be considered a</p><p>sell signal. (Ensure that the ISI is accompanied by price action to</p><p>minimise the possibility that the ISI is just gravitating back to a more</p><p>neutral value.)</p><p>An upward crossOver from the Oversold region on below average or</p><p>low relative volume could be a potential consolidation in price action in</p><p>an existing downtrend.</p><p>A downward crossOver from the Overbought region on below average</p><p>or low relative volume could be a potential consolidation in price action</p><p>in an existing</p><p>uptrend.</p><p>Table 3.1 may help you put ISI values in perspective. You will notice that the</p><p>table is the same as shown for the relative strength index, which is to be</p><p>expected. If you can appreciate the degree of bias required to create an</p><p>extreme value, you should also be able to understand why indicators such as</p><p>the RSI and ISI seldom spend excessive time in the more extreme regions of</p><p>indicator value and why, once they have reached these extreme values, they</p><p>return to the more neutral region of 50 per cent fairly promptly. This should</p><p>also help you understand why you should be very hesitant to buy a stock that</p><p>has an extremely high indicator value.</p><p>Applying the ISI to price action</p><p>My principal application for the ISI is locating divergences with trending</p><p>stocks. I will also examine the ISI for extreme values when considering</p><p>position entry as I do not want to enter a stock that is experiencing a</p><p>historically high ISI value if the trend is reasonably mature. The likelihood of</p><p>ISI values staying in extreme regions for extended periods is low, which</p><p>makes this indicator particularly receptive to relative application. Under</p><p>normal circumstances, I do not act solely on crossOvers from extended</p><p>regions — I will use these occurrences as a weight of evidence when</p><p>assessing current stock activity. However, having experienced some success</p><p>with adapting the RSI to price action, adapting the ISI to price action was a</p><p>natural progression of the Overall concept. (I will not repeat the process for</p><p>the adaptation, having already cOvered this in chapter 2.)</p><p>Figure 3.2 shows a chart of EQI with the ISI adapted to price action. As</p><p>previously discussed with the RSI, figure 3.2 shows that the peaks and</p><p>troughs in ISI indicator development coincided with the prominent points in</p><p>price action, which may lead some to believe that these indicator values are</p><p>ideal. Added to this there is the presence of what many would consider a</p><p>healthy divergence, further confirming underlying weakness.</p><p>Figure 3.2: internal strength index adapted to price action</p><p>Figure 3.3 shows the complete internal strength channel — that is, with</p><p>neutral, Overbought and Oversold regions shaded — using the default 14</p><p>periods. (Again, I will not repeat the process to achieve this, as this was</p><p>cOvered in chapter 2.)</p><p>Figure 3.3: internal strength channel</p><p>As you can see, the internal strength channel is not in sync with trend</p><p>development. As will be discussed in chapter 7, this is largely because the</p><p>default setting of 14 periods has been used to create the index and the</p><p>channel. For now, I’ll continue to use the default 14 periods to create the</p><p>benchmark.</p><p>Establishing a benchmark — internal</p><p>strength channel</p><p>Once again, I will define and test a benchmark that can then be used for</p><p>further comparison as the relative process unfolds.</p><p>Testing parameters</p><p>The testing parameters used to test the internal strength channel concept and</p><p>so create the benchmark for further development are as follows:</p><p>channel periods — 14</p><p>entry trigger — a close above the lower cord of the bullish band</p><p>exit trigger — a close below the lower cord of the bullish band or a</p><p>close below the upper cord of the bullish band for price action that has</p><p>moved into the Overbought region.</p><p>Test results</p><p>The test results for the internal strength channel benchmark, also shown in</p><p>table 3.2, are as follows:</p><p>Table 3.2: internal strength channel — benchmark</p><p>Profit summary</p><p>Profit status PROFITABLE</p><p>Starting capital $25 000.00</p><p>Finishing capital $58 087.69</p><p>Maximum equity/(date) $33 087.69 (30/09/2005)</p><p>Minimum equity/(date) −$5794.18 (29/12/2000)</p><p>Gross trade profit $73 770.58 (295.08%)</p><p>Gross trade loss −$40 682.89 (−162.73%)</p><p>Total net profit $33 087.69 (132.35%)</p><p>Average profit per trade $133.96</p><p>Profit factor 1.8133</p><p>Profit index 44.85%</p><p>Total transaction cost $9880.00</p><p>Total slippage $0.00</p><p>Daily compound interest rate 0.0403%</p><p>Annualised compound interest rate 15.8383%</p><p>Trades processed 3998</p><p>Trades taken 247</p><p>Partial trades taken 0</p><p>Trades rejected 3751</p><p>Winning trades 78 (31.58%)</p><p>Losing trades 169 (68.42%)</p><p>Largest winning trade/(date) $6889.62 (24/05/2002)</p><p>Largest losing trade/(date) −$1230.47 (13/12/2001)</p><p>Average winning trade $945.78</p><p>Average losing trade −$240.73</p><p>Average win/average loss 3.9288</p><p>Trade duration statistics (all trades)</p><p>Maximum trade duration 322 (days)</p><p>Minimum trade duration 5 (days)</p><p>Average trade duration 49 (days)</p><p>Winning trades</p><p>Maximum trade duration 322 (days)</p><p>Minimum trade duration 7 (days)</p><p>Average trade duration 110 (days)</p><p>Losing trades</p><p>Maximum trade duration 128 (days)</p><p>Minimum trade duration 5 (days)</p><p>Average trade duration 22 (days)</p><p>Consecutive trade statistics</p><p>Maximum consecutive winning trades 4</p><p>Maximum consecutive losing trades 12</p><p>Average consecutive winning trades 1.39</p><p>Average consecutive losing trades 3.02</p><p>Relative drawdown</p><p>Maximum dollar drawdown/(date) $3090.78 (3/03/2000)</p><p>Maximum percentage drawdown/(date) 12.400% (3/03/2000)</p><p>Absolute (peak to valley) dollar drawdown</p><p>Maximum dollar drawdown $5834.18 (23.3400%)</p><p>Capital peak/(date)1 $25 000.00 (18000101)</p><p>Capital valley/(date) $19 165.82 (29/12/2000)</p><p>A success rate of 31.58 per cent is comparable with the test results</p><p>achieved with the RSI benchmark, which is to be expected. Like the RSI</p><p>benchmark, this rate is unacceptable.</p><p>The number of opportunities that have been identified, at 3998, is again</p><p>excessive. Wading through this many potential trades is too ridiculous to</p><p>consider, and another previous problem has recurred.</p><p>The largest loss is $1230.47, which is starting to push the boundaries of</p><p>becoming unacceptable. Losses of this magnitude are not an option when</p><p>a trading account starts at only $25 000. While this also highlights a</p><p>problem with simulated trading — as, in real time, most people would</p><p>have closed the position well before the situation deteriorated to this</p><p>extent — for the purposes of continuity in testing and so that all</p><p>benchmarks are exposed to the same set of conditions, I have not</p><p>interfered with the initial process.</p><p>The peak to valley drawdown of 23.34 per cent is unacceptable —</p><p>although, as mentioned, this is a worst-case scenario. It is surprising just</p><p>how many people will stew Over the peak to valley drawdown while</p><p>disregarding all other information. Sure, the peak to valley is as important</p><p>as the other information included; however, it is no more important. I</p><p>prefer to focus on averages as this is closer to real-time expectancy.</p><p>The average loss is $240.73, which is more than acceptable.</p><p>The average net profit is $945.78, which is again comparable to that</p><p>achieved in chapter 2 (table 2.2) with the relative price channel</p><p>benchmark.</p><p>The annualised rate of return is acceptable at 15.84 per cent. However,</p><p>although this is better than bank interest, it falls short of the return</p><p>produced by the relative price channel benchmark.</p><p>As expected the results are somewhat similar to the relative price channel.</p><p>These results show that the benchmarks could still be improved.</p><p>Equity curve and yearly profits</p><p>An examination of the closed trade equity graph shown in figure 3.4</p><p>highlights that the bulk of profits were generated during 2002 and 2003,</p><p>although it can be assumed that the profits in 2002 come from a limited</p><p>number of trades.</p><p>Figure 3.4: closed equity curve for 14-period internal strength channel</p><p>While I disagree with the straight line theory for a rising equity curve, as it</p><p>indicates a system that potentially cuts strong trades short, an equity curve</p><p>such as shown here is potentially unacceptable for real-time application.</p><p>While the testing results indicated an average return of 15.83 per cent Over</p><p>the testing period, in reality three acceptable years compensated for the less</p><p>than satisfactory performance during 2004 and 2005 and the shocker in 2000.</p><p>As discussed at the start of this chapter, the ISI is not smoothed, so it is</p><p>referencing trend development Over the same period as the RSI but without</p><p>manipulation. While the equity curve here is less than ideal, in my opinion it</p><p>is</p><p>an improvement Over the equity curve for the relative price channel, as</p><p>shown in chapter 2 (figure 2.12). Never assume that the inclusion of</p><p>smoothing must improve the end result. The success rate improved to 31.58</p><p>per cent, up from the RSI result of 23.98 per cent.</p><p>Figure 3.5 shows the yearly profits achieved by the benchmark.</p><p>Figure 3.5: yearly profits for the 14-period relative strength channel</p><p>Looking at the performance broken down into yearly blocks, how would you</p><p>have felt during 2004 if you were using this concept? The market boomed</p><p>and you would have gone nowhere; however, you would be pretty chuffed</p><p>with the outcome from 2001 to 2003. The results are reasonably consistent</p><p>considering that no smoothing exists in the current format.</p><p>What is disappointing is that a benchmark that uses what is basically a raw</p><p>version of the RSI has produced a more consistent result than the benchmark</p><p>that used the RSI. While the net return distinctly favoured the RSI, the ISI</p><p>was comparable when analysed more completely. While the smoothing</p><p>potentially increased the net return, it came with a severe price.</p><p>Are you starting to question people’s blind and unquestioned faith in the</p><p>application of indicator parameters based on accepted practice? So far I have</p><p>tested two of the most popular indicators used in confirmation analysis at</p><p>values generally accepted as being suitable for position trading, and produced</p><p>less than satisfactory results. However, many people bank on their indicators</p><p>being right more often than not, without ever assessing whether this is</p><p>actually the case.</p><p>Coding for chapter 3</p><p>Internal strength index — BullCharts</p><p>[Target=Percent; author=Wilson, Leon]</p><p>Periods:=input("Periods", 34, 1, 250);</p><p>Smoothing:=input("Smoothing", 1, 1, 55);</p><p>Trigger:=input("Trigger Line", 1, 1, 55);</p><p>Value2:=input("Over Bought", 70, 50, 99);</p><p>Value3:=input("Over Sold", 30, 1, 50);</p><p>Value4:=input("Upper Neutral Zone", 55, 50, 99);</p><p>Value5:=input("Lower Neutral Zone", 45, 1, 50);</p><p>UpDay:= sum(max(CLOSE - Ref(CLOSE,-1),0),Periods);</p><p>DownDay:= sum(max(Ref(CLOSE,-1) - CLOSE,0), Periods);</p><p>Bias:=100-(100/(1+(UpDay/(DownDay+0.0001))));</p><p>Index:= Ma(Bias,Smoothing,E);</p><p>Trigg:=Ma(Ma(Bias,Smoothing,E),Trigger,E);</p><p>[Color=Black;Linestyle=Solid]</p><p>Index;</p><p>[Color=Blue;Linestyle=dash]</p><p>Trigg;</p><p>[Color=Light Slate Gray;Linestyle=Dotted]</p><p>value3;</p><p>value2;</p><p>[name=1</p><p>st</p><p>Zone1]</p><p>Value5;</p><p>[name=2</p><p>nd</p><p>Zone2]</p><p>value4;</p><p>[name=Bullish Zone; linestyle=Fill;color=Light Slate Gray]</p><p>Value2; Value4;</p><p>[name=Neutral Zone; linestyle=Fill;color=White]</p><p>value5; Value4;</p><p>[name=Bullish Zone; linestyle=Fill;color=Dark Grey]</p><p>Value3; Value5;</p><p>{Copyright Leon Wilson, 2005}</p><p>Internal strength channel — BullCharts</p><p>[Target=Price; author=Wilson, Leon]</p><p>Periods:=input("Periods", 34, 1, 250);</p><p>Smoothing:=input("Smoothing", 1, 1, 55);</p><p>Value2:=input("Over Bought", 70, 51, 99);</p><p>Value3:=input("Over Sold", 30, 1, 49);</p><p>Value4:=input("Upper Neutral Zone", 55, 50, 99);</p><p>Value5:=input("Lower Neutral Zone", 45, 1, 50);</p><p>UpDay := sum(max(CLOSE - Ref(CLOSE,-1),0),Periods);</p><p>DownDay := sum(max(Ref(CLOSE,-1) - CLOSE,0),Periods);</p><p>Index:=100-(100/(1+(UpDay/(DownDay+0.0001))));</p><p>[Color=Light Slate Gray;Linestyle=Dotted]</p><p>OB:=Ma((Index -Value3),Smoothing,E);</p><p>OS:=Ma((Index -Value2),Smoothing,E);</p><p>NZU:=Ma((Index -Value4),Smoothing,E);</p><p>NZL:=Ma((Index -Value5),Smoothing,E);</p><p>{Show Results}</p><p>[name=OverB;]</p><p>Close-(Close*(OB/100));</p><p>[name=OverS;]</p><p>Close-(Close*(OS/100));</p><p>[name=NeutUp;]</p><p>Close-(Close*(NZU/100));</p><p>[name=NeutLower;]</p><p>Close-(Close*(NZL/100));</p><p>{Upper Channel}</p><p>[name=Upper; linestyle=Fill;color=Light Slate Gray]</p><p>Close-(Close*(NZU/100)); Close-(Close*(OS/100));</p><p>{Neutral Zone}</p><p>[name=Zone; linestyle=Fill;color=White]</p><p>Close-(Close*(NZU/100)); Close-(Close*(NZL/100));</p><p>{Lower Channel}</p><p>[name=Lower ; linestyle=Fill;color=Dark Grey]</p><p>Close-(Close*(NZL/100));Close-(Close*(OB/100));</p><p>{Copyright Leon Wilson, 2005}</p><p>Internal strength index — MetaStock</p><p>Periods:=Input("Periods", 1, 250, 34);</p><p>Smoothing:=Input("Smoothing", 1, 55, 1);</p><p>Value2:=Input("Over Bought",50, 99, 70);</p><p>Value3:=Input("Over Sold", 1, 50, 30);</p><p>Value4:=Input("Upper Neutral Zone", 50, 99, 55);</p><p>Value5:=Input("Lower Neutral Zone", 1,50, 45);</p><p>UpDay:= Sum(Max(CLOSE - Ref(CLOSE,-1),0),Periods);</p><p>DownDay:= Sum(Max(Ref(CLOSE,-1) - CLOSE,0),Periods);</p><p>Bias:=100-(100/(1+(UpDay/(DownDay+0.0001))));</p><p>Index:= Mov(Bias, Smoothing,E);</p><p>Trigg:=Mov(Mov(Bias,Smoothing,E),1,E);</p><p>Index; Trigg; Value3; Value2; Value5; Value4;</p><p>{Copyright - Leon Wilson, 2005}</p><p>Internal Strength Channel - MetaStock</p><p>Periods:=Input("Periods", 1, 250, 34);</p><p>Smoothing:=Input("Smoothing", 1, 55, 1);</p><p>Value2:=Input("Over Bought", 50, 99, 70);</p><p>Value3:=Input("OverSold", 1, 50, 30);</p><p>Value4:=Input("Upper Neutral Zone", 50, 99, 55);</p><p>Value5:=Input("Lower Neutral Zone", 1, 50, 45);</p><p>UpDay := Sum(Max(CLOSE - Ref(CLOSE,-1),0),Periods);</p><p>DownDay := Sum(Max(Ref(CLOSE,-1) - CLOSE,0),Periods);</p><p>Index:=100-(100/(1+(UpDay/(DownDay+0.0001))));</p><p>OB:=Mov((Index -Value3),Smoothing,E);</p><p>OS:=Mov((Index -Value2),Smoothing,E);</p><p>NZU:=Mov((Index -Value4),Smoothing,E);</p><p>NZL:=Mov((Index -Value5),Smoothing,E);</p><p>CLOSE-(CLOSE*(OB/100));</p><p>CLOSE-(CLOSE*(OS/100));</p><p>CLOSE-(CLOSE*(NZU/100));</p><p>CLOSE-(CLOSE*(NZL/100));</p><p>{Copyright Leon Wilson, 2005}</p><p>Internal strength channel — TradeSim (MetaStock)</p><p>Pds:=14;</p><p>SU := Sum(Max(CLOSE - Ref(CLOSE,-1),0),Pds);</p><p>SD := Sum(Max(Ref(CLOSE,-1) - CLOSE,0),Pds);</p><p>Index:=100-(100/(1+(SU/(SD+0.0001))));</p><p>NZU:=Index -55;</p><p>NZL:=Index -45;</p><p>UC:=Index -70;</p><p>NeutUp:=CLOSE-(CLOSE*(NZU/100));</p><p>OB:=CLOSE-(CLOSE*(UC/100));</p><p>NeutL:= CLOSE-(CLOSE*(NZL/100));</p><p>EntryTrigger := (CLOSE>NeutUp) AND (Ref(CLOSE,-1)<=Ref(NeutUp,-1)) ;</p><p>EntryPrice := OPEN;</p><p>ExitTrigger := (CLOSE<NeutUp AND (Ref(CLOSE,-1) >=Ref(NeutUp,-1))) OR</p><p>(CLOSE<OB AND (Ref(CLOSE,-1) >=Ref(OB,-1))) ;</p><p>ExitPrice := OPEN;</p><p>InitialStop:=0;</p><p>{Copyright Leon Wilson, 2005}</p><p>1 As can be seen in table 3.2, the capital peak entry and the date listed seem</p><p>to be incorrect; however, the values shown are correct. In order to expose</p><p>this strategy (and all strategies tested throughout this book) to the highest</p><p>probability of failure, I always commence testing from a market high or</p><p>mature phase in development. This usually ensures that the strategy is</p><p>exposed to the greatest and most consistent losses first up. In other words,</p><p>the strategy was running at a loss from day 1. I intentionally commenced</p><p>testing from a point where the highest probability of failure existed. If you</p><p>lose money from day one, could you have survived long enough for things</p><p>to improve? While many traders will commence testing a system in the</p><p>early run of a bullish market, in reality the majority of newcomers start</p><p>dabbling in the market near trend maturity. This means they commence</p><p>trading with minimal capital in a mature market and then have to survive</p><p>through the next bear market while surrendering the bulk of their existing</p><p>capital. Hence, the potential gain from testing is inconsistent with actual</p><p>behaviour. I aim to replicate this behaviour by commencing the testing</p><p>process at the point where losses are the highest probability. The results are</p><p>not as good, but I feel they are more realistic.</p><p>Chapter 4</p><p>The final frontier (well, almost)</p><p>Not even close actually — I have barely begun. So far I have discussed the</p><p>adaptation of a conventional index to price action, then the adaptation of an</p><p>oscillator to price action, which led to the inclusion of the internal strength</p><p>index. This chapter will look at the adaptation of directional movement to</p><p>price action. Some may consider this to be far too complex; however, this is</p><p>not the case. I will start from the beginning, as you must understand the</p><p>process if you are to have any comprehension of the final result.</p><p>Directional movement</p><p>Directional movement is one of the few confirmation tools that works as well</p><p>on the right side of the screen as it does on the left. As with all indicators,</p><p>directional movement is not perfect nor is it the Holy Grail of confirmation</p><p>tools, and I</p><p>is not a familiar world. Any similarities to</p><p>past experiences of buying and selling are strictly limited to the surface —</p><p>the world underneath remains forever foreign to the majority.</p><p>The belief that trading is easy stems from people’s everyday actions when</p><p>buying and selling unwanted items for a few dollars. True story. My mate had</p><p>an old lawnmower that was worth absolutely nothing to the average person;</p><p>however, he managed to sell it for $70. I called in one Sunday afternoon for a</p><p>barbecue and he proudly told me how he managed to sell that dreadful old</p><p>mower for $70 (which in turn he wisely invested in two cartons of beer),</p><p>when he would have been happy with $20. He viewed this as a $50 profit.</p><p>Many people carry the same mentality into the market, whereby they</p><p>believe that the ability to sell an old fridge that is worth basically nothing for</p><p>a few dollars more than it is really worth translates into the ability to sell</p><p>shares successfully. Nothing could be further from the truth. The truth is that</p><p>my mate accepted a $400 loss rather than $450. When most people sell</p><p>second-hand items, they forget that the sale price is discounted to the original</p><p>purchase price. The new price of the mower was $470 not $20, so no profit</p><p>ever existed — my mate merely reduced the impact of the loss by selling at a</p><p>slightly better price. What the majority fail to realise when trading is that they</p><p>always buy at a new price and then must sell at an even higher price than the</p><p>initial cost in order to profit.</p><p>In other words, they have to set a new benchmark high selling price. This is</p><p>often unkindly referred to as the ‘greater fool theory’ — that is, there’s a</p><p>greater fool than me. If people traded mowers instead of shares, in order to</p><p>make a profit, my mate would have been required to sell the mower at more</p><p>than the initial purchase price of $470. In order to actually generate $50</p><p>profit, the sale price would have to have been $520. Unfortunately, the</p><p>concept of buying at a new price and selling at an even higher price escapes</p><p>the majority. The trick is successfully identifying the ‘greatest fools of all’ —</p><p>that is, those who are buying at the top of the market. I aim to address the</p><p>common problems associated with identifying mature and unsustainable price</p><p>action throughout this book.</p><p>The discussions in this book are directed toward the position trader, using a</p><p>relatively conservative mindset; however, the concept that I have developed</p><p>can be satisfactorily applied to literally all trading time frames. The</p><p>application of this concept is potentially so vast and its impact so significant</p><p>that, once it is fully understood, it may cause many people to reassess how</p><p>they trade from a number of perspectives. The application of this technique is</p><p>potentially so diverse that it is impossible for me to cOver it in exhaustive</p><p>detail here. With this in mind, my approach has been directed more toward</p><p>the introduction of the concept rather than in-depth specifics involving</p><p>relative analysis and its adaptation to price action. If I can explain to you the</p><p>concept of relative adaptive analysis successfully, you have the building</p><p>blocks for ongoing personal development.</p><p>The learning curve usually develops in small steps, with many traders not</p><p>realising the evolutionary process in which they participate until one day they</p><p>sit back and think about issues with an understanding that was previously</p><p>lacking. While the bulk of progress is usually subtle, occasionally some of us</p><p>experience a monumental leap in application — and this is what this book is</p><p>about.</p><p>Some will ask, ‘Why share something potentially so significant?’ My</p><p>response is, ‘Why not?’ Those people who ask this sort of question still</p><p>believe that the answer to successful trading lies solely within one indicator,</p><p>or that there is a single correct solution. If this is your thought process, you</p><p>are probably not ready for this book. The indicator is only one piece of a</p><p>much larger process and to think along these lines reflects a level of trading</p><p>immaturity. Some traders consider that their personal indicator is so super</p><p>special that they dare not share the coding, yet if they took the time to look in</p><p>the indicator menu on programs such as MetaStock and BullCharts, there is a</p><p>distinct possibility that they will find something comparable reasonably</p><p>quickly.</p><p>Mature and robust trading evolves in unison with the progression of the</p><p>successful trader’s understanding of his or her environment. Successful</p><p>traders thrive on change and welcome the presence of a constant challenge,</p><p>while others are happy to trade using proven techniques, accepting time-</p><p>honoured practice at face value. As you will discOver by the end of this</p><p>book, I am not one of the latter group of traders. I constantly probe the</p><p>boundaries of technical analysis and question what I have been told from a</p><p>practical perspective, regardless of who the source may be, and refuse to</p><p>accept anything at face value until I am satisfied that what I am told is correct</p><p>or I can disprove the concept relative to my own approach and understanding.</p><p>While many inexperienced traders chase that one illusive indicator that will</p><p>give them the definitive answer they seek, they fail to study trend</p><p>development. I attempt to convey the concept and value of learning price</p><p>action as a primary part of trading survival; however, very few seem willing</p><p>to listen. Why? I suspect that it does not accommodate the ‘must have now’</p><p>mentality that has become so entrenched in our society, nor does it offer a</p><p>single definitive solution or provide the level of certainty they seek. The</p><p>majority are prepared to learn how to trade as long as they have it mastered</p><p>by this afternoon and the process is in black and white. When I or other</p><p>seasoned traders are unable to accomplish this, we are looked upon as falling</p><p>short of being useful or, worse still, deliberately being unhelpful. We are</p><p>viewed as being reluctant to share the real secrets. Never for one moment is it</p><p>considered that the expectations may be unrealistic.</p><p>Indicators in their classic format have always been a bone of contention</p><p>with me. The majority of indicators when applied at their default values on a</p><p>daily time frame are basically useless. As I have repeatedly suggested to my</p><p>subscribers, indicators on daily are primarily a confidence tool rather than an</p><p>instrument for confirmation, and repeated testing tends to bear this out.</p><p>However, I need to clarify one point when it comes to indicators — it’s not</p><p>the concept of an indicator that concerns me, but rather the method of</p><p>application.</p><p>For years traders have plotted indicators on the bottom of the charts in a</p><p>separate window using 14 periods, and this has been done without question.</p><p>From this they endeavour to draw a conclusion relative to trend development,</p><p>assuming that the indicator at the bottom of the chart is sufficiently reflective</p><p>of trend development to enable an enhanced decision through its application.</p><p>Why hasn’t someone taken indicators and adapted them to price action, so</p><p>that an indicator such as the relative strength index has a monetary value</p><p>rather then a percentage value? Wouldn’t it be better to show Overbought</p><p>regions on trend development rather than in a separate pane? I had all of these</p><p>questions and no answers. Gradually Over time I worked out how to convert</p><p>the relative strength index back to a dollar value. I always considered that this</p><p>was a distinct possibility as the bulk of indicators originate from price action</p><p>in the first place. If the indicator commences with a monetary value, usually</p><p>the closing price, it was merely a process of converting the percentage value</p><p>back to its original format without undermining the integrity of the indicator.</p><p>I have commenced with the adaptation of the relative strength index to</p><p>price action for no other reason than it’s probably the most widely used</p><p>indicator in existence today. In chapter 3, I will continue with the internal</p><p>strength index and then, in chapter 4, I will move</p><p>am not going to pretend otherwise. Like all indicators, it has its</p><p>faults and shortcomings. This is not a problem if you take the time to</p><p>understand the strengths and weaknesses of your trading tools.</p><p>One of the biggest problems with conventional directional movement is</p><p>that the technique can be indecisive in trend development. Directional</p><p>movement is divided into two elements, as follows:</p><p>positive directional movement — PDI or DI+</p><p>negative directional movement — MDI or DI−</p><p>As each of these elements of directional movement evaluates bullish and</p><p>bearish price action independently, there is no direct comparison of the</p><p>relationship that exists between the two entities. While you may be able to</p><p>see that positive directional movement is the dominant indicator, you won’t</p><p>know the degree to which it dominates negative directional movement.</p><p>If the relationship between the two driving forces behind price action can</p><p>be quantified, then the degree of bias favouring one element of price action</p><p>Over the other, and whether this element has become an unsustainable</p><p>influence, can be determined. The more extreme bullish or bearish activity</p><p>becomes, the more unsustainable the move will be from a longer term</p><p>perspective. Opportunists will see a use for an end to extreme sentiment</p><p>regardless of whether it is motivated by the bulls or the bears.</p><p>In order to quantify the disparity between the bulls and the bears, I set</p><p>about developing the directional index. One of the primary problems with the</p><p>conventional indicator is the lack of continuity — regarding the crossOver</p><p>from bullish to bearish, for example. Depending on the values of each</p><p>element of price action, the crossOver of the DI indicators can occur at</p><p>different values, leading to a lack of consistency with regard to this element</p><p>of the analysis process. Added to this, another question arises. Let’s consider</p><p>the following. Let’s say that the DI+ indicator has a value of 40 per cent,</p><p>while the DI– indicator has a value of 10 per cent. This would indicate a very</p><p>bullish presence underlying trend development. If the DI+ continued to</p><p>produce a value of 40 per cent and the DI–rose to a value of 15 per cent, there</p><p>would be a 50 per cent increase in the bearish element underlying trend</p><p>development with the same bullish value. The visual change in this</p><p>relationship is usually minimal at best.</p><p>The problem here is that the two relationships described above differ yet it</p><p>is difficult to accurately quantify the change in the underlying relationship.</p><p>To gain a more complete picture of directional movement, ideally the two</p><p>indicators need to be combined. As good as the original indicator may be, the</p><p>difficulty is in knowing where, for example, the bullish bias finishes and the</p><p>bearish bias starts. The directional index effectively defines the relationship</p><p>that exists between the DI+ and the DI–. The default indicator highlights the</p><p>underlying strength and weakness simultaneously while the directional index</p><p>defines the bias within theses two forces driving trend development. Never</p><p>underestimate the power of knowing the bias that lies within trend</p><p>development.</p><p>The directional index is a direct comparison between positive and negative</p><p>directional movement Over a specified number of periods while consistently</p><p>defining the crossOver from a bullish to a bearish dominance. You can see</p><p>where I’m going with this, can’t you?</p><p>Figure 4.1 shows the directional index, which puts indicator behaviour into</p><p>context.</p><p>Figure 4.1: conventional directional movement and directional index</p><p>Directional index values</p><p>Directional index values reveal the following points:</p><p>An index rising from a low value but with a value less than 50 per cent</p><p>reveals that the bears are dominant but the bearish sentiment is</p><p>weakening.</p><p>A rising index above 50 per cent indicates that the bulls are dominant</p><p>and are gaining in strength.</p><p>A declining index from an extremely high value reveals that the bulls</p><p>are still dominant but the bullish sentiment is weakening.</p><p>A declining index with a value less than 50 per cent indicates that the</p><p>bears are in control and are gaining in strength.</p><p>The same values and percentage table can be applied as previously discussed</p><p>for the internal strength index and similar range-bound indicators such as the</p><p>relative strength index.</p><p>Hence, when analysing the directional index, the following guidelines can</p><p>be applied:</p><p>A value of 50 per cent indicates equal positive and negative sentiment.</p><p>For immature uptrends, values greater than 70 per cent indicate the early</p><p>onset of extreme bullish behaviour and the potential for up-trending</p><p>activity.</p><p>For mature uptrends, values greater than 70 per cent indicate extreme</p><p>bullish behaviour and that the uptrend is becoming unsustainable.</p><p>For immature downtrends, values less than 30 per cent indicate the early</p><p>onset of extreme bearish behaviour and the potential for down-trending</p><p>activity.</p><p>For mature downtrends, values less than 30 per cent indicate extreme</p><p>bearish behaviour and that the downtrend is potentially becoming</p><p>unsustainable.</p><p>One observation that I have made Over time is that uptrends will seldom</p><p>develop on neutral indicator values; however, down-trending activity can</p><p>develop without experiencing excessive bearish indicator support. This</p><p>means you should use the above as a guide only.</p><p>The directional index and trading</p><p>The directional index values can be used in the following ways, depending on</p><p>whether you are a position or a short-term trader.</p><p>Position trading:</p><p>Entry — an upward crossOver of the index above the value of 50 per</p><p>cent would be considered a confirmation signal.</p><p>Exit — a downward crossOver below the value of 50 per cent would be</p><p>accepted as a confirmation signal.</p><p>Short-term trading:</p><p>Entry — an upward crossOver of the index above the trigger line would</p><p>be considered a confirmation signal.</p><p>Exit — a downward crossOver of the index below the trigger line would</p><p>be considered a confirmation signal.</p><p>Question</p><p>Go back to figure 4.1 on page 87. Do you think you could put a dollar value on the bullish</p><p>activity of indicator development marked on the chart. That is, in dollar terms, just how</p><p>dominant is bullish price action? Have a go before you read on.</p><p>Directional index on price</p><p>When I originally designed the directional index, it was not intended to</p><p>replace the more conventional approach to DI analysis; however, its</p><p>adaptation into index format and so to price action has improved the</p><p>technique from my perspective. This index allows you to quantify extreme</p><p>bullish or bearish behaviour and the bias or balance of power within the</p><p>market. It has all of the characteristics of normal indices and allows for</p><p>divergence analysis. It will show a continued bias toward extreme bullish</p><p>values in up-trending conditions while highlighting a bias of extreme bearish</p><p>values in down-trending stocks. This is a characteristic common with most</p><p>range-bound indices such as the RSI, relative volatility index, money flow</p><p>index and others. The directional index behaves in the same manner as all</p><p>indicators when applied to illiquid and spasmodically traded stocks.</p><p>The adaptation of directional movement into an index format meant that</p><p>extreme points in trend development could be quantified. This single step has</p><p>also opened up the opportunity of applying the directional movement to price</p><p>action. Furthermore, converting the indicator value to a monetary value</p><p>means that the relationship between sustainable directional movement and</p><p>ongoing trend development can be directly correlated.</p><p>Figure 4.2 shows the directional index on price.</p><p>Figure 4.2: directional movement on price</p><p>Before proceeding, I need to put some minds at ease with regard to the</p><p>adaptation of directional movement to price action and that it can actually be</p><p>done without pages of complex calculations. While it is possible to include</p><p>the DI+ and DI– indicators on price action, I prefer to include a line that</p><p>reflects the degree of bias, or favouritism to look at it another way, that is</p><p>underpinning</p><p>trend development. This is the value as defined by the</p><p>directional index, so this is the value that must be adapted to price action. The</p><p>conventional directional movement indicator at the bottom of the chart</p><p>defines both bullish and bearish activity, and which dominant component is</p><p>driving trend development. Nothing new here you say. If you refer back to</p><p>figure 4.1, you will see a question mark at the point where, later, I asked you</p><p>to have a go at defining a dollar value for the bullish presence that is</p><p>underlying price action. As figure 4.2 shows, the bullish dominance at the</p><p>time equated to $0.59; therefore, for the bearish component of price action to</p><p>become the dominant force the next day, a change of approximately $0.59</p><p>would have been required. Naturally, it’s not quite that straightforward — a</p><p>bearish move would also draw the directional movement line toward</p><p>declining price action, thereby reducing the separation value. What is of</p><p>importance here, however, is that the approximate change needed in price for</p><p>the trend to take on an opposing dominance can be defined. In this case, the</p><p>directional index is $0.59 below price action, so this can be used to determine</p><p>the likelihood of the stock experiencing a bearish crossOver. On a stock that</p><p>is trading at $1.60, is highly unlikely that the bears will be taking charge</p><p>tomorrow.</p><p>I feel fairly confident in saying that the majority of you would have</p><p>considered this an impossibility not that long ago — like, a few pages back.</p><p>On figure 4.2, I have also highlighted the crossOver points with conventional</p><p>directional movement. This confirms that the crossOver of price action is</p><p>consistent with conventional indicator behaviour. When I ended up with</p><p>directional movement on price action, I was like a kid with a new toy. The</p><p>mind went into Overdrive — as one would expect from a trader who holds</p><p>in-depth discussions about market behaviour with a rather round and</p><p>somewhat uninterested Jack Russell.</p><p>Figure 4.3 shows the full directional channel.</p><p>Figure 4.3: directional price channel at 14 periods</p><p>Figure 4.3 again shows that, once an indicator is adapted to price, it has little</p><p>relevance if 14 periods are used. By now you should be beginning to</p><p>appreciate the consequence of applying indicators that appear relative to price</p><p>action on the surface but when adapted to price action their relationship is</p><p>shown to be questionable at best. The use of values that are more relative to</p><p>price action will be discussed in chapter 7.</p><p>Establishing the benchmark</p><p>If directional movement can be applied to price action, then extreme regions</p><p>of trend development relative to directional movement can also be defined. I</p><p>have applied my directional price, or momentum bias, channel to price action</p><p>using the market accepted 14 periods. The same problems encountered with</p><p>the RSI and ISI concepts can immediately be seen. The indicator is a</p><p>complete mismatch with regard to trend development. However, before I start</p><p>exploring the world of more compatible parameters, I will discuss how the</p><p>default channel performs at 14 periods.</p><p>I have continued with the process of creating a benchmark result using the</p><p>same parameters as previously.</p><p>Testing parameters</p><p>The testing parameters used to create the benchmark are as follows:</p><p>channel periods — 14</p><p>entry trigger — a close above the lower cord of the bullish band</p><p>exit trigger — a close below the lower cord of the bullish band or a</p><p>close below the upper cord of the bullish band for price action that has</p><p>moved into the Overbought region.</p><p>Test results</p><p>The test results for the momentum bias channel benchmark, as shown in table</p><p>4.1, are as follows:</p><p>Table 4.1: momentum bias channel — benchmark</p><p>Profit summary</p><p>Profit status PROFITABLE</p><p>Starting capital $25 000.00</p><p>Finishing capital $112 246.14</p><p>Maximum equity/(date) $91 572.43 (29/04/2005)</p><p>Minimum equity/(date) −$6668.64 (23/11/2001)</p><p>Gross trade profit $135 631.32 (542.53%)</p><p>Gross trade loss −$48 385.19 (−193.54%)</p><p>Total net profit $87 246.14 (348.98%)</p><p>Average profit per trade $398.38</p><p>Profit factor 2.8032</p><p>Profit index 64.33%</p><p>Total transaction cost $8760.00</p><p>Total slippage $0.00</p><p>Daily compound interest rate 0.0723%</p><p>Annualised compound interest rate 30.1695%</p><p>Trades processed 2863</p><p>Trades taken 219</p><p>Partial trades taken 0</p><p>Trades rejected 2644</p><p>Winning trades 66 (30.14%)</p><p>Losing trades 153 (69.86%)</p><p>Largest winning trade/(date) $37 102.56 (7/05/2004)</p><p>Largest losing trade/(date) −$2363.00 (9/05/2003)</p><p>Average winning trade $2055.02</p><p>Average losing trade −$316.24</p><p>Average win/average loss 6.4982</p><p>Trade duration statistics (all trades)</p><p>Maximum trade duration 842 (days)</p><p>Minimum trade duration 5 (days)</p><p>Average trade duration 70 (days)</p><p>Winning trades</p><p>Maximum trade duration 842 (days)</p><p>Minimum trade duration 6 (days)</p><p>Average trade duration 159 (days)</p><p>Losing trades</p><p>Maximum trade duration 140 (days)</p><p>Minimum trade duration 5 (days)</p><p>Average trade duration 32 (days)</p><p>Consecutive trade statistics</p><p>Maximum consecutive winning trades 4</p><p>Maximum consecutive losing trades 16</p><p>Average consecutive winning trades 1.50</p><p>Average consecutive losing trades 3.40</p><p>Relative drawdown</p><p>Maximum dollar drawdown/(date) $4173.02 (1/08/2003)</p><p>Maximum percentage drawdown/(date) 12.7600% (14/04/2000)</p><p>Absolute (peak to valley) dollar drawdown</p><p>Maximum dollar drawdown $6708.64 (26.8300%)</p><p>Capital peak/(date) $25 000.00 (18000101)</p><p>Capital valley/(date) $18 291.36 (23/11/2001)</p><p>The number of possible trades continues to be excessive at 2863. While</p><p>this is a significant reduction when compared with the other concepts</p><p>tested so far, it remains far from satisfactory.</p><p>The success rate has just scraped above 30 per cent, with the number of</p><p>successful trades coming to 30.14 per cent. This is nothing to get excited</p><p>about just yet.</p><p>The average win/loss is 6.49. This is actually not too shabby for an</p><p>approach that has been tested solely on the closing price crossing above</p><p>the lower cord of the bullish band. This is not scientific by any stretch of</p><p>the imagination, so to produce a win/loss ratio on this scale should not be</p><p>underestimated.</p><p>The one result that most people will be interested in is the annualised</p><p>return and, at 30.169 per cent, you would have to be pretty happy this</p><p>result, especially as this is the bench mark for further improvement.</p><p>The average profit was a respectable $2055.02, while the average loss</p><p>was $316.24.</p><p>The peak to valley drawdown is unacceptable at 26.83 per cent.</p><p>Naturally, you would need to see a reduction in this value as the concept</p><p>is adapted to price action.</p><p>Again I will not pore Over every detail produced by the testing process;</p><p>however, the concept of applying directional movement to price action does</p><p>appear to be the superior concept so far in the initial phase of understanding</p><p>relative analysis. At this point, the indicator values have only been converted</p><p>from percentage to monetary in order to allow for adaptation to price. The</p><p>next step is to adjust the channels so they are relative to trend development.</p><p>Equity curve and yearly profits</p><p>Figure 4.4 shows the equity curve for the momentum bias channel</p><p>benchmark.</p><p>Figure 4.4: closed equity curve for the 14-period momentum bias</p><p>channel</p><p>I’m the first to admit that this is an ordinary looking equity curve; however,</p><p>at least its general direction since the commencement of the bull run in March</p><p>2003 is upward, although it has plateaued near the end of the testing period.</p><p>The general upward direction that coincides with the bull run is encouraging</p><p>— at least there is something to work with. Figure 4.5 shows the yearly</p><p>profits for the momentum bias channel.</p><p>Figure 4.5: yearly profits for the 14-period momentum bias channel</p><p>As can be seen from figure 4.5, the bulk of profits were gained in 2004.</p><p>While this is not surprising given the strength of the market, 2003 did not</p><p>perform that well. On the other hand, 2001 was a reasonably flat year for the</p><p>market, so to produce a positive result of this magnitude in bearish conditions</p><p>should not be underestimated — although this</p><p>has probably stemmed from a</p><p>couple of sizeable trades.</p><p>A sizeable loss of 18.94 per cent was incurred in 2000, which on its own</p><p>would not be a good start. Adding this to the erratic nature of returns since</p><p>would make this a difficult tool to trust in real time.</p><p>Once again, it is important to remember that at this point no additional</p><p>methods of confirmation or means of filtering potentially genuine from false</p><p>signals have been included, so to come out in front when the general market</p><p>is in decline is more than satisfactory as a starting point. What this does</p><p>underline is the poor performance of the concept in 2003, which proved to</p><p>the commencement of a long bullish run. This is a potential weakness that</p><p>will require some attention if this concept is to become truly relative to trend</p><p>development.</p><p>Conclusions so far</p><p>What is becoming increasingly apparent is the inconsistent nature of</p><p>indicators when applied at default values. Even the more highly respected</p><p>indicators are producing ordinary results. It really doesn’t matter which</p><p>calculation process is used — the success rate is low and the general manner</p><p>in which the equity curve develops is often erratic. The yearly returns are</p><p>generally spasmodic and the losses often severe and sustained. The</p><p>disconcerting aspect of this research is that the majority of traders unwittingly</p><p>apply these values in position trading, in the belief that confirmation at these</p><p>parameters is actually enhancing their performance. In reality, it’s probably</p><p>having the opposite effect. Whether you agree with what has been discussed</p><p>so far or not, if you are a realist the only conclusion you can practically draw</p><p>is that the relationship between confirmation analysis and trend development</p><p>is not relative. There is price action and there are the preferred tools of</p><p>analysis, but from the perspective of a position trader there is no tangible</p><p>relationship between the two.</p><p>Coding for chapter 4</p><p>Directional index — BullCharts</p><p>[Target=Percent; author=Wilson, Leon]</p><p>Periods:=input("DI Periods", 34, 1, 250);</p><p>Smoothing:=input("Smoothing", 1, 1, 55);</p><p>Trigger:=input("Trigger Line", 1, 1, 55);</p><p>Value2:=input("Over Bought", 70, 50, 99);</p><p>Value3:=input("Over Sold", 30, 1, 50);</p><p>Value4:=input("Upper Neutral Zone", 55, 50, 99);</p><p>Value5:=input("Lower Neutral Zone", 45, 1, 50);</p><p>up := PDI(Periods);</p><p>down := MDI(Periods);</p><p>Index := 100-(100/(1+(up/down)));</p><p>[Color=Black;Linestyle=Solid]</p><p>Ma(Index,Smoothing,E);</p><p>[Color=Blue;Linestyle=Dash]</p><p>Ma(Ma(Index,Smoothing,E),Trigger,E);</p><p>[Color=Light Slate Gray;Linestyle=Dotted]</p><p>value3; value2;</p><p>[name=1</p><p>st</p><p>Zone1]</p><p>Value5;</p><p>[name=2</p><p>nd</p><p>Zone2]</p><p>value4;</p><p>[name=Bullish Zone; linestyle=Fill;color=Light Slate Gray]</p><p>Value2; Value4;</p><p>[name=Neutral Zone; linestyle=Fill;color=White]</p><p>value5; Value4;</p><p>[name=Bullish Zone; linestyle=Fill;color=Dark Grey]</p><p>Value3; Value5;</p><p>{Copyright Leon Wilson, 2005}</p><p>Directional price channel — BullCharts</p><p>[Target=Price; author=Wilson, Leon]</p><p>Periods:=input("DI Periods", 34, 1, 250);</p><p>Smoothing:=input("Smoothing", 1, 1, 55);</p><p>Trigger:=input("Trigger Line", 1, 1, 55);</p><p>Value2:=input("Over Bought", 70, 50, 99);</p><p>Value3:=input("Over Sold", 30, 1, 50);</p><p>Value4:=input("Upper Neutral Zone", 55, 50, 99);</p><p>Value5:=input("Lower Neutral Zone", 45, 1, 50);</p><p>up := PDI(Periods);</p><p>down := MDI(Periods);</p><p>Index := 100-(100/(1+(up/down)));</p><p>[Color=Light Slate Gray;Linestyle=Dotted]</p><p>OB:=Ma((Index -Value3),Smoothing,E);</p><p>OS:=Ma((Index -Value2),Smoothing,E);</p><p>NZU:=Ma((Index -Value4),Smoothing,E);</p><p>NZL:=Ma((Index -Value5),Smoothing,E);</p><p>{Show Results}</p><p>[name=OverB;]</p><p>Close-(Close*(OB/100));</p><p>[name=OverS;]</p><p>Close-(Close*(OS/100));</p><p>[name=NeutUp;]</p><p>Close-(Close*(NZU/100));</p><p>[name=NeutLower;]</p><p>Close-(Close*(NZL/100));</p><p>{Upper Channel}</p><p>[name=Upper; linestyle=Fill;color=Light Slate Gray]</p><p>Close-(Close*(NZU/100)); Close-(Close*(OS/100));</p><p>{Neutral Zone}</p><p>[name=Zone; linestyle=Fill;color=White]</p><p>Close-(Close*(NZU/100)); Close-(Close*(NZL/100));</p><p>{Lower Channel}</p><p>[name=Lower ; linestyle=Fill;color=Dark Grey]</p><p>Close-(Close*(NZL/100));Close-(Close*(OB/100));</p><p>{Copyright Leon Wilson, 2005}</p><p>Directional index — MetaStock</p><p>Periods:=Input("DI Periods", 1, 250, 34);</p><p>Smoothing:=Input("Smoothing", 1, 55, 1);</p><p>Value2:=Input("Over Bought", 50, 99, 70);</p><p>Value3:=Input("Over Sold", 1, 50, 30);</p><p>Value4:=Input("Upper Neutral Zone", 50, 99, 55);</p><p>Value5:=Input("Lower Neutral Zone", 1, 50, 45);</p><p>up := PDI(Periods);</p><p>down := MDI(Periods);</p><p>Index := 100-(100/(1+(up/down)));</p><p>Mov(Index,Smoothing,E);</p><p>Mov(Mov(Index,Smoothing,E),1,E);</p><p>Value3; Value2; Value5; Value4;</p><p>{Copyright Leon Wilson, 2005}</p><p>Directional price channel — MetaStock</p><p>Periods:=Input("DI Periods", 1,250, 34);</p><p>Smoothing:=Input("Smoothing", 1, 51, 1);</p><p>Value2:=Input("Over Bought", 50, 99, 70);</p><p>Value3:=Input("Over Sold", 1, 50, 30);</p><p>Value4:=Input("Upper Neutral Zone", 50, 99, 55);</p><p>Value5:=Input("Lower Neutral Zone", 1, 50, 45);</p><p>up := PDI(Periods);</p><p>down := MDI(Periods);</p><p>Index := 100-(100/(1+(up/down)));</p><p>OB:=Mov((Index -Value3),Smoothing,E);</p><p>OS:=Mov((Index -Value2),Smoothing,E);</p><p>NZU:=Mov((Index -Value4),Smoothing,E);</p><p>NZL:=Mov((Index -Value5),Smoothing,E);</p><p>CLOSE-(CLOSE*(OB/100));</p><p>CLOSE-(CLOSE*(OS/100));</p><p>CLOSE-(CLOSE*(NZU/100));</p><p>CLOSE-(CLOSE*(NZL/100));</p><p>{Copyright Leon Wilson, 2005}</p><p>Directional price channel — TradeSim (MetaStock)</p><p>Pds:=14;</p><p>Index:=100-(100/(1+(PDI(Pds)/(MDI(Pds)+0.0001))));</p><p>NZU:=Index -55;</p><p>NZL:=Index -45;</p><p>UC:=Index -70;</p><p>OB:=CLOSE-(CLOSE*(UC/100));</p><p>NeutUp:=CLOSE-(CLOSE*(NZU/100));</p><p>NeutL:= CLOSE-(CLOSE*(NZL/100));</p><p>EntryTrigger := CLOSE>NeutUp AND (Ref(CLOSE,-1)<=Ref(NeutUp,-1));</p><p>EntryPrice := OPEN;</p><p>ExitTrigger := (CLOSE<NeutUp AND (Ref(CLOSE,-1) >=Ref(NeutUp,-1))) OR</p><p>(CLOSE<OB AND (Ref(CLOSE,-1) >=Ref(OB,-1))) ;</p><p>ExitPrice := OPEN;</p><p>InitialStop:=0;</p><p>{Copyright Leon Wilson, 2005}</p><p>Chapter 5</p><p>All things relative</p><p>So far the focus has been on the relationship between price action and</p><p>indicator values using price channels. This is an internal assessment of price</p><p>action itself. The relative aspect of the price channels occurs on two fronts:</p><p>The bias that exists between bullish and bearish activity. The more I</p><p>think about it, I tend to believe that this is probably where Wilder</p><p>derived the ‘relative’ aspect of the relative strength index. While it is not</p><p>relative to any external influences, the indicator displays the relationship</p><p>that exists between bullish and bearish activity. The indicator value is a</p><p>representation of the bias associated with whichever element of price</p><p>action is dominating.</p><p>Price channels relative to general trend development. While the</p><p>indicators discussed so far may display the relationship between higher</p><p>and lower closes, they do not define the relationship relative to general</p><p>trend development. Ultimately, the parameters that are eventually</p><p>applied to the price channels in this book are moulded to coincide with</p><p>general trend activity, thereby encompassing bullish and bearish activity</p><p>from a relative perspective.</p><p>Is it possible to evaluate and, more importantly, display market strength in</p><p>the same manner? Let’s say you wanted to compare a stock with the general</p><p>market. I tend to focus on strength when long side trading, especially in a</p><p>weak or bearish climate; however, this relationship is often difficult to</p><p>quantify. At best, there is the garden variety indicator at the bottom of the</p><p>screen, which usually tells you very little. For every trader who understands</p><p>what the indicator represents, ten more do not. The concept of relative</p><p>strength comparison (RSC) is fine, but the way in which this relationship is</p><p>projected has left a lot to the imagination. The desire to create an index that</p><p>more accurately defined individual stock behaviour against the actions of the</p><p>market in general led to the development of the comparative strength index.</p><p>Before looking at this, I will briefly discus the RSC.</p><p>Relative strength comparison</p><p>Let’s go back to the basics for one moment. I will use my version of the</p><p>relative strength index in order to create the comparative</p><p>strength index. As</p><p>already discussed, the RSI evaluates bullish and bearish activity</p><p>independently, and then displays the relationship between the two; however,</p><p>this style of indicator can be adapted for comparative analysis. The standard</p><p>RSC tool for most programs is usually a single day ratio of the relationship</p><p>that exists between two securities. More often than not, it is defined as</p><p>follows:</p><p>Stock/Security</p><p>Not exactly complex, is it? I will quickly revisit the conventional approach to</p><p>comparative strength analysis by applying the technique to National Australia</p><p>Bank and the finance sector index (XFJ), as shown in figure 5.1.</p><p>Figure 5.1: conventional strength analysis</p><p>The indicator at the bottom of the chart on figure 5.1 is the default RSC tool</p><p>as most commonly applied. The latest closing price for the National Australia</p><p>Bank was $31.94 while the finance sector index closed at 5427.80 points. The</p><p>value shown for the relationship that exists between the finance sector index</p><p>and NAB — that is, $31.94 divided by 5427.80 — is 0.006, which tells you</p><p>absolutely nothing.</p><p>As BullCharts rounds to three decimal places, a value of 0.006 appears on</p><p>the chart, as shown in figure 5.1. MetaStock will usually display the full</p><p>value as calculated. While the final value has little meaning, the capacity of</p><p>trading software to define the relationship that exists between multiple</p><p>securities as a single value allows you to rectify the majority of analytical</p><p>shortcomings that crop up when utilising multiple security analysis.</p><p>The general approach to interpreting the default RSC indicator as it appears at</p><p>the bottom of figure 5.1 is as follows:</p><p>A rising indicator value suggests that the stock is outperforming the</p><p>index.</p><p>A declining indicator value suggests that the stock is underperforming</p><p>the index.</p><p>A non-trending indicator value suggests that the stock mirrors the index</p><p>in general behaviour.</p><p>Comparative strength index</p><p>The RSC is a fairly rudimentary approach that on its own doesn’t really tell</p><p>you a great deal, other than the disparity that exists between two closing</p><p>prices. It does not tell you anything about the sustainability of the relationship</p><p>nor does it identify any points of extreme and potentially unsustainable</p><p>behaviour. Thankfully, the calculation process allows you to define the</p><p>relationship that exists between two securities as a single value, which in turn</p><p>permits you to Overcome this shortcoming. Indeed, the original RSI can be</p><p>converted into a comparison indicator, meaning that the relationship that</p><p>exists between the stock and index can be more accurately defined. This</p><p>adaptation is what created the comparative strength index. As already shown</p><p>in chapter 2, the coding for the RSI is as follows:</p><p>Periods:=input("Calculation Periods", 14, 1, 250);</p><p>Range:=(2*Periods)-1;</p><p>UpDay:=ma(If(CLOSE>Ref(CLOSE,-1),CLOSE-Ref(CLOSE,-1),0),Range,E);</p><p>DownDay:=ma(If(CLOSE<Ref(CLOSE,-1),Ref(CLOSE,-1)-CLOSE,0),Range,E);</p><p>100-(100/(1+(UpDay/(DownDay+0.000001))))</p><p>I have commenced by selecting an indicator that has a single field of trading</p><p>data, as shown above. By this I am referring to a formula that only requires</p><p>one piece of data in order to complete its calculation process. In the case of</p><p>the RSI, you will notice that the entire calculation is based on the closing</p><p>price and nothing else. In order to convert a conventional indicator into a</p><p>comparison indicator, you simply substitute the closing price with the</p><p>calculation process used by the conventional RSC indicator, as shown below:</p><p>Ratio:=Close/Security;</p><p>Periods:=input("Calculation Periods", 14, 1, 250);</p><p>Range:=(2*Periods)-1;</p><p>UpDay:=ma(If(Ratio>Ref(Ratio,-1), Ratio-Ref(Ratio,-1),0),Range,E);</p><p>DownDay:=ma(If(Ratio<Ref(Ratio,-1),Ref(Ratio,-1)-Ratio,0),Range,E);</p><p>100-(100/(1+(UpDay/(DownDay+0.000001))))</p><p>You will notice that the closing price no longer exists and that it has been</p><p>replaced by the term ‘Ratio’. The reference to ratio calculates the disparity</p><p>that exists between the two securities under analysis in the same manner as</p><p>the conventional relative strength comparison indicator. The coding now</p><p>calculates separately the days where the relationship between the two closing</p><p>prices has either expanded or contracted, in the same manner as the RSI</p><p>calculates the daily fluctuation of a stock’s closing price. The coding then</p><p>displays this relationship as an index, again in the same manner, thus creating</p><p>the comparative strength index (CSI). At last something decent to work with.</p><p>Please be aware that this type of calculation process is usually limited to</p><p>those indicators that require a single field of data — such as the RSI or the</p><p>ISI, where the closing price is the only field of data required. More complex</p><p>indicators such as the money flow index or directional movement, which</p><p>require multiple fields of data — such as the closing price and volume or the</p><p>high, low and closing price used in the stochastic, for example — will not</p><p>function correctly. While these problems can be Overcome, an advanced</p><p>knowledge of practical coding and of the individual formulas concerned is</p><p>required in order to address this issue successfully. I will not cOver this here</p><p>as it is going well beyond our intended level of discussion.</p><p>Applying the CSI to trading software</p><p>So far the above calculation process merely highlights the change in coding</p><p>in order to create a comparative index; however, the above coding will not</p><p>function on its own when applied to trading software. As I am sure that you</p><p>will explore this approach in more detail, probably using your own indicators,</p><p>I will briefly explain how to activate the calculation process using BullCharts</p><p>and MetaStock. With both programs you need to include an additional two</p><p>lines of code in order to trigger the calculation process. These are outlined</p><p>below.</p><p>BullCharts</p><p>First up, you need to locate the index to include in order to commence the</p><p>calculation process. This is coded as follows:</p><p>Symbol:=inputsymbol("Security","XJO");</p><p>This tells BullCharts to locate the required index. In this case the S&P/ASX</p><p>200 Index (XJO) is set at default. You can change this as needed when you</p><p>apply the CSI to your chart.</p><p>The second and final line of code tells BullCharts to divide the closing</p><p>price of the stock by the closing price of the selected index, as shown below:</p><p>Ratio:=Close/LoadSymbol(symbol, Close);</p><p>From here the calculation process occurs as above.</p><p>MetaStock</p><p>MetaStock users need to apply a slightly different approach in order to</p><p>achieve the same result. With BullCharts, you can simply select the required</p><p>index from the database; however, with MetaStock you need to define the</p><p>actual directory path in order to locate the required security. The coding for</p><p>this instruction is usually as follows:</p><p>Symbol:=Security("C:\My Databases\MetaStock\ASX\X0\XJO",CLOSE);</p><p>This line tells MetaStock to follow the specified directory path and to select</p><p>the S&P/ASX 200 Index (XJO). Once MetaStock locates the appropriate</p><p>security, it is to reference the closing price for calculation purposes, as shown</p><p>below:</p><p>Ratio:=CLOSE/Symbol;</p><p>This tells MetaStock to divide the closing price of the stock and the closing</p><p>price of the selected index.</p><p>Don’t panic if it’s all becoming a bit too confusing — it will become</p><p>clearer as you enter the coding into your trading program. The only problem</p><p>for MetaStock users is that the actual directory path will vary depending upon</p><p>individual data suppliers and the designated hard drive used for data storage</p><p>by the user. Should you encounter problems with identifying the correct</p><p>directory path, please contact your data supplier for technical support as I am</p><p>unable to help with directory issues specific to individual data vendors. The</p><p>above directory path is compatible with the default JustData installation.</p><p>Behaviour and values of the CSI</p><p>Before you can start displaying indicators on charts that highlight the</p><p>relationship that exists between two securities, you need to have some idea of</p><p>the situations that will cause the CSI</p><p>to fluctuate as the relationship evolves</p><p>— remember, however, that the following points are guidelines only.</p><p>If the CSI is rising:</p><p>the stock is rising at a rate greater than the index</p><p>the stock is declining at a lesser rate than the index</p><p>the stock is non-trending while the index is declining</p><p>the stock is rising while the index is non-trending.</p><p>If the CSI is declining:</p><p>the stock is declining at a rate greater than the index</p><p>the stock is rising at a lesser rate than the index</p><p>the stock is non-trending while the index is rising</p><p>the stock is declining while the index is non-trending.</p><p>If the CSI is non-trending, the change in the stock is comparable to the</p><p>change experienced by the index.</p><p>Table 5.1 shows the value and relevant percentages for the CSI. (I have</p><p>included the table even though it is the same as previous tables so you are not</p><p>required to reference earlier chapters.) The CSI defines the relationship</p><p>between two securities; it does not define the strength or behaviour of the</p><p>individual security. In order to more fully appreciate this particular concept</p><p>you first need to understand the relationship that you are attempting to</p><p>evaluate. As a rule of thumb, you can apply the following as a basic guide for</p><p>interpretation:</p><p>Table 5.1: values and percentages for comparative strength index</p><p>Indicator value % Strength</p><p>0 100.0% (bearish)</p><p>10 80.0%</p><p>20 60.0%</p><p>30 40.0%</p><p>40 20.0%</p><p>50 0.0% (neutral)</p><p>60 20.0%</p><p>70 40.0%</p><p>80 60.0%</p><p>90 80.0%</p><p>100 100.0% (bullish)</p><p>A value of 50 per cent indicates that the change experienced by the stock</p><p>is comparable to the change experienced by the index Over the same</p><p>period.</p><p>A value of 70 per cent reflects a 40 per cent bullish bias in favour of the</p><p>stock compared to the index Over the same time period.</p><p>A value of 30 per cent reflects a 40 per cent bearish bias in favour of the</p><p>stock compared to the index Over the same time period.</p><p>You will notice that I have not included the usual references to rising or</p><p>falling stocks. Remember — strong bullish or bearish values can occur while</p><p>one of the securities is non-trending. A strong indicator merely defines a</p><p>situation where the stock is behaving in a manner contrary to the index. It</p><p>does not automatically imply that the stock is trending.</p><p>The values shown in table 5.1 may lead people to question how such</p><p>conclusions are drawn. The same principles apply here as previously applied</p><p>to the RSI, ISI and directional movement, whereby bullish and bearish</p><p>activity are defined separately as a part of the calculation process. It’s this</p><p>relationship between the two forces that determines indicator value. With a</p><p>conventional approach, you can reasonably associate strong indicator values</p><p>with trending activity; however, this is not the case with comparative</p><p>analysis. While the same percentage values may be occurring, the assumption</p><p>that trending activity exists is not always correct.</p><p>Figure 5.2 shows both stock and index on a single chart, which is nothing</p><p>new to anyone who has been trading for a while. The need occasionally arises</p><p>for two securities to be Overlaid in order to evaluate the relationship between</p><p>the two entities. In some cases, this can be achieved with some effect;</p><p>however, more often than not it is difficult to accomplish more than a general</p><p>assessment, or to define the relationship with any degree of accuracy. It may</p><p>show whether the stock or index is performing better, but it does not show the</p><p>precise relationship that exists between the two.</p><p>Figure 5.2: comparative strength index</p><p>The comparative strength index does exactly that. Not only do you have the</p><p>ability to define the relationship between two securities, this association can</p><p>also be quantified. Added to this, you can also gauge when outlandish</p><p>activity is becoming unsustainable — for example, stocks, especially the</p><p>larger blue chips, will not be allowed to run excessively or to pull too far</p><p>away from the market average before profits are collected.</p><p>The same principles as discussed in chapter 2 apply, whereby the CSI also</p><p>defines any change at twice its actual occurrence — that is, a 10 per cent</p><p>change in indicator values represents a 5 per cent change in the underlying</p><p>relationship. Knowing this allows you to quantify market strength Over a</p><p>specific period. Let’s examine figure 5.2 in more detail.</p><p>The following areas have been highlighted in figure 5.2:</p><p>A EquiGold was drifting in a non-trending direction at this point while the</p><p>index declined during the same period. This produced a bullish CSI as</p><p>price action, by refusing to trade in a similar downward manner, was</p><p>stronger than the general market. As mentioned, a bullish indicator does</p><p>not always imply that a stock is trending, as you would expect with a</p><p>conventional indicator. The strength in this scenario was not in trending</p><p>activity but rather the unwillingness of the market to sell EQI lower in</p><p>unison with the broader market.</p><p>B EQI became parabolic, rocketing skyward while, during the same</p><p>period, the index also lifted. However, the CSI is still rising because the</p><p>stock has risen at rate greater than the index. Many look at strength</p><p>analysis and say, ‘So what? It’s outperforming the index.’ This is an</p><p>incorrect and a somewhat simplistic view of the actions of an index. On</p><p>average, EQI is outperforming the top 200 stocks. If I were to say to you,</p><p>‘Here is a stock that is outperforming general market activity’, I would</p><p>most likely have your undivided attention; however, the moment I tender</p><p>the same information but put it into the context of outperforming an</p><p>index, the majority immediately lose interest. The problem is that the</p><p>index is viewed in the same manner as a single stock when this is not</p><p>what it represents.</p><p>C At this point, EQI immediately plateaus following the parabolic rally;</p><p>however, the index continues its steady march toward higher values. This</p><p>promptly causes the CSI to decline. Remembering that the CSI defines the</p><p>relationship between the closing prices of two securities, you can now</p><p>define exactly how EQI is performing compared with general market</p><p>behaviour. This indicator tells you that there are stronger stocks in the</p><p>market, simply because it is in decline.</p><p>There was a brief period where the XJO and EQI rose at a comparable</p><p>rate and the CSI drifted at a neutral value.</p><p>In the boxed region of price action, both the index and EQI have declined;</p><p>however, the CSI shows that EQI has been sold down excessively when</p><p>compared with general market behaviour. This is indicated by the CSI</p><p>dropping into the underperforming region.</p><p>Comparative strength channel</p><p>Now for the really exciting stuff. Having devised a process that quantifies the</p><p>relationship between two securities — for example, EQI and the XJO — I</p><p>could now place a comparative strength channel on price action. How good is</p><p>this? At last broad market strength could be applied to a stock. I must admit</p><p>that when I accomplished this for the first time — that is, I could directly</p><p>quantify broad market activity against the price action of a lone security using</p><p>a single scale and have the value actually mean something — it was fairly</p><p>exciting. The adaptation of broad market strength to a single security in a</p><p>manner that was visually beneficial rates right up there with the first bream</p><p>run of the season. Now that is exciting stuff. For those readers who like a spot</p><p>of bream fishing, I am talking about fish consistently in excess of 40</p><p>centimetres in length and pushing three kilograms. They are built like</p><p>miniature tree stumps and prone to busting light tackle with monotonous</p><p>regularity. Perhaps you are wondering what my bream fishing has to do with</p><p>trading. It’s about understanding your environment. You will never conquer</p><p>your adversary if you do not understand what you are up against.</p><p>Figure 5.3 shows the adaptation of the comparative strength index to price</p><p>and the comparative strength channel.</p><p>Figure 5.3: price and market comparison</p><p>While I am reluctant to claim it as a first, I would dare say that very few</p><p>people apply comparative analysis</p><p>in the same manner as I do. Not only do I</p><p>perform relative analysis, I conduct this from a comparative perspective.</p><p>The following areas have been highlighted in figure 5.3:</p><p>1 Price action that pushes beyond the upper cord of the bullish band is</p><p>outperforming the market to the point of becoming unsustainable when</p><p>compared with surrounding broad market activity. For the first time the</p><p>term ‘Overbought’ is probably justified.</p><p>As the comparative strength channel is relative to price action while</p><p>defining average broad market behaviour as defined by those stocks from</p><p>which the major index is derived, it can reasonably be assumed that when</p><p>a stock breaches the upper cord of the bullish band it has genuinely</p><p>become Overbought. Try to remember that if you are analysing a large</p><p>blue chip such as ANZ Bank using the comparative strength channel, its</p><p>behaviour directly affects the very channel that you are using for analysis.</p><p>If the stock directly contributes to channel development but does not</p><p>reflect channel behaviour — for example, through producing an extreme</p><p>value such as defined by the Overbought region — the stock has the</p><p>potential to be exactly that, Overbought when compared with surrounding</p><p>stocks that define the major index. If a particular stock has been</p><p>excessively punished in a sell-off or, conversely, a buying frenzy has</p><p>occurred, the market will capitalise on the lower trading price or collect</p><p>profits. Genuine Overbought and Oversold conditions can only be defined</p><p>when the behaviour of one stock is compared with the actions of many.</p><p>For the first time, there is the potential to identify such points in trend</p><p>behaviour through the implementation of price channels.</p><p>2 Price action that develops within the confines of the bullish band shows</p><p>that, on average, the stock is outperforming the broad market.</p><p>3 Price action that develops within the neutral region of the comparative</p><p>strength channel indicates that the behaviour of the stock and that of the</p><p>index are comparable, with the degree of change similar for both.</p><p>4 When price action drops into the bearish band of the comparative</p><p>strength channel, the stock is underperforming the index and the broad</p><p>market on average.</p><p>5 Price action that drops below the lower cord of the bearish band has</p><p>potentially been sold down excessively when compared with the general</p><p>market. Remember — the bearish activity of EQI is compared with 200</p><p>other stocks as determined by the S&P/ASX 200 index. When EQI drops</p><p>below the lower cord it has been treated far more severely than</p><p>surrounding stocks, so for the first time there is a legitimate case for the</p><p>stock being Oversold.</p><p>The benefits of applying comparative strength directly to price action are too</p><p>numerous to mention here; however, for the first time you have the ability to</p><p>trade effectively using comparative strength. Added to this, you can now</p><p>develop both long and short side strategies based around comparative relative</p><p>strength and then trade in accordance with the strength or weakness</p><p>underpinning trend development. When price action drops into the bearish</p><p>band of the comparative strength channel, the stock in question is, on</p><p>average, underperforming the broader market, so a short side strategy could</p><p>be applied. Likewise, you can trade in accordance with strength by limiting</p><p>long side strategies to price action that develops within the confines of the</p><p>bullish band of the comparative strength channel.</p><p>Establishing a benchmark</p><p>Naturally, the question arises as to whether you could trade using the</p><p>comparative strength channel as a stand-alone technique. In order to test this,</p><p>I have largely continued in the same manner as in previous chapters for the</p><p>purpose of continuity.</p><p>Testing parameters</p><p>The testing parameters used to create the benchmark are as follows:</p><p>channel periods — 34</p><p>entry trigger — a close above the lower cord of the bullish band</p><p>exit trigger — a close below the upper cord of the bearish band or a</p><p>close below the upper cord of the bullish band for price action that has</p><p>moved into the Overbought region.</p><p>The only way this benchmark differs from those in the previous chapters is</p><p>that I have decided to use 34 periods. I chose not to include a 14-period</p><p>benchmark because the stock is being referenced against the major index</p><p>(XJO). For the large percentage of stocks, this is very similar to referencing</p><p>the stock against the actions of itself. The benchmark is testing the big stocks</p><p>against the S&P/ASX 200 index; however, the index is derived from the very</p><p>stocks being tested against it. As most stocks tend to mirror the index — as</p><p>they must otherwise the index could not behave in the manner it does — a</p><p>short-term comparative strength channel serves only limited benefit in</p><p>creating a realistic benchmark. Hence, I commenced with a 34-period</p><p>comparative strength channel in order to produce a longer term assessment of</p><p>how stocks perform against broad market activity.</p><p>Test results</p><p>If there was ever a classic example of people Over-complicating the process</p><p>of trading and why it is often unnecessary then here it is. The above concept,</p><p>which simply involved acting on a close above the lower cord of the bullish</p><p>band and exiting on basic exit criteria, produced an annualised return of</p><p>20.49 per cent with a profitability rate of 52.38 per cent. The majority of</p><p>people reinvent the wheel from a squillion different perspectives and still</p><p>cannot come close to this degree of consistency. I have not included any</p><p>filtering mechanism in the way of minimum volume or price change. The</p><p>entry signal is based solely on the occurrence of the stock outperforming the</p><p>index while the exit focuses either on extreme and unsustainable</p><p>outperformance or sign of underperformance. I’m not suggesting that this is</p><p>the perfect trading strategy; however, incorporating this style of analysis has</p><p>improved the probability of success. While this analysis may sacrifice</p><p>additional profit, it should improve consistency if the following results are</p><p>anything to go by.</p><p>The test results for the comparative strength channel benchmark, also</p><p>shown in table 5.2, are as follows:</p><p>Table 5.2: comparative strength channel — 34-period benchmark</p><p>Profit summary</p><p>Profit status PROFITABLE</p><p>Starting capital $25 000.00</p><p>Finishing capital $72 051.70</p><p>Maximum equity/(date) $47 964.46 (26/08/2005)</p><p>Minimum equity/(date) −$5 137.75 (2/02/2001)</p><p>Gross trade profit $67 127.10 (268.51%)</p><p>Gross trade loss −$20 075.40 (−80.30%)</p><p>Total net profit $47 051.70 (188.21%)</p><p>Average profit per trade $1120.28</p><p>Profit factor 3.3437</p><p>Profit index 70.09%</p><p>Total transaction cost $1680.00</p><p>Total slippage $0.00</p><p>Daily compound interest rate 0.0511%</p><p>Annualised compound interest rate 20.4982%</p><p>Trades processed 445</p><p>Trades taken 42</p><p>Partial trades taken 0</p><p>Trades rejected 403</p><p>Winning trades 22 (52.38%)</p><p>Losing trades 20 (47.62%)</p><p>Largest winning trade/(date) $9403.84 (11/02/2005)</p><p>Largest losing trade/(date) −$1880.40 (27/05/2005)</p><p>Average winning trade $3051.23</p><p>Average losing trade −$1003.77</p><p>Average win/average loss 3.0398</p><p>Trade duration statistics (all trades)</p><p>Maximum trade duration 1379 (days)</p><p>Minimum trade duration 35 (days)</p><p>Average trade duration 336 (days)</p><p>Winning trades</p><p>Maximum trade duration 1379 (days)</p><p>Minimum trade duration 70 (days)</p><p>Average trade duration 450 (days)</p><p>Losing trades</p><p>Maximum trade duration 616 (days)</p><p>Minimum trade duration 35 (days)</p><p>Average trade duration 211 (days)</p><p>Consecutive trade statistics</p><p>Maximum consecutive winning trades 5</p><p>Maximum consecutive losing trades 4</p><p>Average consecutive winning trades 2.75</p><p>Average consecutive losing trades 2.22</p><p>Relative drawdown</p><p>Maximum dollar drawdown/(date) $5197.75 (2/02/2001)</p><p>Maximum percentage drawdown/(date) 20.7900% (2/02/2001)</p><p>Absolute (peak to valley) dollar drawdown</p><p>Maximum dollar drawdown $5197.75 (20.7900%)</p><p>Capital peak/(date) $25 000.00 (18000101)</p><p>Capital valley/(date) $19 802.25 (2/02/2001)</p><p>The average loss is sizeable at $1003.77.</p><p>The average profit is very impressive at $3051.23.</p><p>The size of the peak to valley drawdown, at 20.70</p><p>per cent would have</p><p>to be considered as a borderline relative drawdown. While a drawdown of</p><p>this magnitude is an undesirable result, the damage caused by the worst</p><p>string of losses was only $5197.75.</p><p>The number of trades processed was very pleasing with 445</p><p>opportunities identified. This is a very manageable result. Not only has</p><p>the success rate broken the 50 per cent barrier, you would not be required</p><p>to wade through an excessive number of trades in search of high</p><p>probability candidates.</p><p>This has also potentially disproved two common misconceptions:</p><p>A low success rate system is required if you intend to produce a decent</p><p>return. While the above results may seem sedate in the context of the</p><p>ongoing discussions, an annualised return of around 20 per cent with a</p><p>success rate of just Over 50 per cent on average is not too shabby for an</p><p>approach that incorporates no filtering</p><p>You must enter as near to the low as possible. The above approach,</p><p>where strength comparison is the sole focus, is probably the slowest</p><p>approach to identifying a point of probable trend entry; however, it</p><p>proved very consistent. The problem with this type of approach is the</p><p>difficulty in identifying consistent entry signals once bullish or bearish</p><p>conditions become established. With the bulk of stocks either under- or</p><p>outperforming once trending conditions become entrenched, the number</p><p>of new candidates tends to be limited. The above benchmark managed to</p><p>generate 445 trading opportunities in total, which is still considerably</p><p>more than most people could trade.</p><p>Equity curve and yearly profits</p><p>The equity curve for the comparative strength channel benchmark, shown in</p><p>figure 5.4, developed in a reasonably consistent manner; however, things</p><p>plateaued after April 2005 (although it is important to remember that testing</p><p>ended on 30 June 2005). The drawdown previously mentioned can easily be</p><p>identified, with the solid decline shown during 2001. There is a second,</p><p>though less significant, drop in capital during March and April 2005. Beyond</p><p>these two points, there is little evidence to suggest that excessive drawdown</p><p>is a major issue.</p><p>Figure 5.4: closed equity curve for 34-period comparative strength</p><p>channel</p><p>Figure 5.5 shows the yearly profits for the comparative strength channel</p><p>benchmark. The yearly breakdown shows a level of consistency that was</p><p>missing in many other strategies at the most basic level. The breakdown is as</p><p>follows:</p><p>Figure 5.5: yearly profits for 34-period comparative strength channel</p><p>2000 — a return of −9.6819 per cent</p><p>2001 — a return of 38.3467 per cent</p><p>2002 — a return of 29.5586 per cent</p><p>2003 — a return of 29.6701 per cent</p><p>2004 — a return of 8.874 per cent</p><p>2005 — a return of 26.1044 per cent.</p><p>With no means of filtering and given the range of market conditions</p><p>experienced, this level of consistency was quite good.</p><p>Coding for chapter 5</p><p>Comparative strength channel — BullCharts</p><p>Ticker:=inputsymbol("Security","XJO");</p><p>Periods:=input("Channel Periods", 34, 1, 250);</p><p>Smoothing:=input("Smoothing", 1, 1, 55);</p><p>Value2:=input("Over Bought", 70, 50, 99);</p><p>Value3:=input("Over Sold", 30, 1, 50);</p><p>Value4:=input("Upper Neutral Zone", 55, 50, 99);</p><p>Value5:=input("Lower Neutral Zone", 45, 1, 50);</p><p>Channel:=Close/LoadSymbol(Ticker, Close);</p><p>[Color=Light Slate Gray;Linestyle=Dotted]</p><p>OB:=Ma((RSI(Channel,Periods)-Value3),Smoothing,E);</p><p>OS:=Ma((RSI(Channel,Periods)-Value2),Smoothing,E);</p><p>NZU:=Ma((RSI(Channel,Periods)-Value4),Smoothing,E);</p><p>NZL:=Ma((RSI(Channel,Periods)-Value5),Smoothing,E);</p><p>{Show Results}</p><p>[name=OverB;]</p><p>Close-(Close*(OB/100));</p><p>[name=OverS;]</p><p>Close-(Close*(OS/100));</p><p>[name=NeutUp;]</p><p>Close-(Close*(NZU/100));</p><p>[name=NeutLower;]</p><p>Close-(Close*(NZL/100));</p><p>{Upper Channel}</p><p>[name=Upper; linestyle=Fill;color=Light Slate Gray]</p><p>Close-(Close*(NZU/100)); Close-(Close*(OS/100));</p><p>{Neutral Zone}</p><p>[name=Zone; linestyle=Fill;color=White]</p><p>Close-(Close*(NZU/100)); Close-(Close*(NZL/100));</p><p>{Lower Channel}</p><p>[name=Lower ; linestyle=Fill;color=Dark Grey]</p><p>Close-(Close*(NZL/100));Close-(Close*(OB/100));</p><p>Comparative strength index — BullCharts</p><p>Symbol:=inputsymbol("Security","XJO");</p><p>Periods:=input("Periods",34,1,250);</p><p>Smoothing:=input("Smoothing", 1, 1, 55);</p><p>Trigger:=input("Trigger Line", 1, 1, 55);</p><p>Value2:=input("Over Bought", 70, 51, 99);</p><p>Value3:=input("Over Sold", 30, 1, 49);</p><p>Value4:=input("Upper Neutral Zone", 55, 50, 99);</p><p>Value5:=input("Lower Neutral Zone", 45, 1, 50);</p><p>Channel:=Close/LoadSymbol(symbol, Close);</p><p>Index:= Ma(RSI(Channel, Periods), Smoothing, E);</p><p>Trigg:=Ma(Ma(RSI(Channel, Periods), Smoothing, E),Trigger,E);</p><p>[Color=Black;Linestyle=Solid]</p><p>Index;</p><p>[Color=Blue;Linestyle=dash]</p><p>Trigg;</p><p>[Color=Light Slate Gray;Linestyle=Dotted]</p><p>value3;value2;</p><p>[name=1st Zone1]</p><p>Value5;</p><p>[name=2nd Zone2]</p><p>value4;</p><p>[name=Bullish Zone; linestyle=Fill;color=Light Slate Gray]</p><p>Value2; Value4;</p><p>[name=Neutral Zone; linestyle=Fill;color=White]</p><p>value5; Value4;</p><p>[name=Bullish Zone; linestyle=Fill;color=Dark Grey]</p><p>Value3; Value5;</p><p>Comparative strength channel — MetaStock</p><p>Periods:=Input("Channel Periods", 1, 250, 34);</p><p>Smoothing:=Input("Smoothing", 1, 55, 1);</p><p>Value2:=Input("Over Bought", 50, 99, 70);</p><p>Value3:=Input("Over Sold", 1, 50, 30);</p><p>Value4:=Input("Upper Neutral Zone", 50, 99, 55);</p><p>Value5:=Input("Lower Neutral Zone", 1, 50, 45);</p><p>BM:=Security("C:\My Databases\MetaStock\ASX\X0\XJO",CLOSE);</p><p>Ratio:=CLOSE/BM;</p><p>Channel:= RSI(Ratio,Periods);</p><p>OB:=Mov((RSI(Channel,Periods)-Value3),Smoothing,E);</p><p>OS:=Mov((RSI(Channel,Periods)-Value2),Smoothing,E);</p><p>NZU:=Mov((RSI(Channel,Periods)-Value4),Smoothing,E);</p><p>NZL:=Mov((RSI(Channel,Periods)-Value5),Smoothing,E);</p><p>CLOSE-(CLOSE*(OB/100));</p><p>CLOSE-(CLOSE*(OS/100));</p><p>CLOSE-(CLOSE*(NZU/100));</p><p>CLOSE-(CLOSE*(NZL/100));</p><p>Comparative strength index — MetaStock</p><p>Periods:=Input("Periods",1,250,34);</p><p>Smoothing:=Input("Smoothing", 1, 55, 1);</p><p>Value2:=Input("Over Bought", 50, 99, 70);</p><p>Value3:=Input("Over Sold", 1, 50, 30);</p><p>Value4:=Input("Upper Neutral Zone", 50, 99, 55);</p><p>Value5:=Input("Lower Neutral Zone", 1, 50, 45);</p><p>BM:=Security("C:\My Databases\MetaStock\ASX\X0\XJO",CLOSE);</p><p>Ratio:=CLOSE/BM;</p><p>Index:= Mov(RSI(Ratio,Periods), Smoothing,E);</p><p>Trigg:=Mov(Mov(RSI(Ratio,Periods),Smoothing,E),1,E);</p><p>Index; Trigg;</p><p>Value3; Value2; Value5; Value4;</p><p>Please note: the above formulas for MetaStock were developed using the</p><p>default JustData directory. As the directory path between data suppliers will</p><p>vary we are unable to include coding suitable for all vendors. Technical</p><p>support should be able assist you with the correct directory path if you</p><p>experience difficulties.</p><p>Chapter 6</p><p>Mingle with the crowd</p><p>While adapting an indicator to price action is a step in the right direction with</p><p>regard to relative analysis, we are far from the finish line. The progression</p><p>from relative strength to the internal strength index was a natural and logical</p><p>progression. From this I moved onto the adaptation of the directional</p><p>movement to price action, and then comparative strength to price action. Now</p><p>I will look at adapting volume to price action. This will be the real surprise</p><p>package, especially for those who have never considered using volume as an</p><p>analysis and strategy tool in the same manner as a momentum indicator.</p><p>So far the bulk of the adaptation I have discussed has involved momentum-</p><p>based indicators such as the ISI and directional movement. While this may</p><p>help in the understanding of trend change, it does not reflect underlying</p><p>market support. Market support is ultimately determined by the number of</p><p>people who are prepared to put capital at risk. How volume reacts with trend</p><p>development can then be analysed.</p><p>Tip</p><p>What do you think would be the most important piece of information — the degree of price</p><p>change or the volume associated with the change in price? Price can change without</p><p>notable volume support; however, trends cannot survive without volume supporting one</p><p>element of price action, regardless of what you may have been told. It’s the simple laws of</p><p>supply and demand.</p><p>Let’s look at two possible</p><p>scenarios and you decide which one has the greater</p><p>significance:</p><p>The closing price finishes in the bullish band for the first time. The</p><p>trading range — that is, the change in price — is consistent with</p><p>previous trend development, while volume is also consistent with</p><p>previous levels of liquidity. The closing price is also the high of the day.</p><p>The closing price finishes in the bullish band for the first time. The</p><p>trading range has increased by more than 5 per cent Over yesterday</p><p>while there has also been a 75 per cent jump in volume. The closing</p><p>price is also the high of the day.</p><p>The answer is fairly obvious. A higher closing price that develops on</p><p>increased trading activity signifies increased market awareness. While the</p><p>question will always remain as to whether this is a genuine signal, increased</p><p>volume is a sign of increased enthusiasm for a particular stock, and with</p><p>increased interest pushing prices up to close on a high, this basic price and</p><p>volume relationship cannot be ignored.</p><p>One of the shortcomings with regards to volume analysis occurs during</p><p>established trend development. Volume is generally heavy at the</p><p>commencement and conclusion of trending activity, which is no coincidence.</p><p>When a trend initial commences and is in the early stages of development,</p><p>volume is usually at its heaviest. The larger private traders, institutional</p><p>investors and those who have a reasonable technique for identifying early</p><p>trend activity will be entering the stock. As price action settles down into an</p><p>established trend, volume then eases. The slightly more conservative traders</p><p>enter the stocks along with the ‘mum and dad’ investors. Occasionally, the</p><p>bigger players may pyramid into a position; however, generally speaking</p><p>volume can become a fairly sedate and indecisive analysis tool as the trend</p><p>rolls on toward maturity.</p><p>At trend conclusion, the larger players begin to lighten their position by</p><p>selling. Again, the astute private traders may also have picked this point to</p><p>exit or may have the ability to pick the actions of the professionals through</p><p>the increased volume activity. As the closing price begins to decrease,</p><p>indicating that supply is outstripping demand, the position is closed usually in</p><p>conjunction with a very specific exit strategy.</p><p>The problem with conventional volume analysis, as with most analysis</p><p>tools, is it may fail to identify any definitive behaviour on occasions. While</p><p>volume may be most obvious at the beginning and conclusion of trend</p><p>development, the six months of trending activity in between can develop</p><p>without displaying significant volume activity or producing any behaviour</p><p>that is obvious enough to grab traders’ attention.</p><p>Volume bias index</p><p>Is it possible to define a relationship between volume activity and trend</p><p>change? Most people would answer no, simply for no other reason than that,</p><p>while volume can fluctuate significantly on a daily basis without producing a</p><p>definitive relationship with general trend development, trends develop with a</p><p>degree of consistency. On the surface, there appears to be no sustained</p><p>correlation between volume and ongoing trend development, so most people</p><p>assume that no definable relationship exists. While volume on its own can be</p><p>as decisive as it can be indecisive, if volume is separated into bullish and</p><p>bearish elements, buying and selling pressure should be able to be reviewed</p><p>simultaneously. If I could do this, I would also be able to define any</p><p>discrepancy in the form of an index. This particular line of thought led to the</p><p>creation of the volume bias index — and I had no idea at the time whether it</p><p>would prove to be beneficial.</p><p>The coding for the volume bias index is as follows:</p><p>UpDay:=(Sum(If(CLOSE>=Ref(CLOSE,-1),VOLUME,0),14))/14;</p><p>DownDay:=(Sum(If(Ref(CLOSE,-1)>=CLOSE,VOLUME,0),14))/14;</p><p>Index:= 100-(100/(1+(Up/Down)));</p><p>As you can see, the basic formula is not that dissimilar to the others discussed</p><p>so far. The first line of the volume bias index tells the program to record the</p><p>level of volume each time the closing price is equal to or higher than previous</p><p>Over 14 periods. If the closing price is not higher than previous then a value</p><p>of zero is recorded. Volume is totalled then averaged Over the same period,</p><p>which eliminates the impact of low liquidity and one-off bumps in trading</p><p>activity. Often the volume associated with a single day’s trading can show</p><p>very little; however, knowing the collective level of trading activity can have</p><p>a significant impact on the decision-making process when you know what</p><p>you are looking for.</p><p>Naturally the process is reversed for lower closes — that is, the volume is</p><p>summed for all closing prices lower than previous and then averaged.</p><p>The disparity between volume associated with higher closes as against</p><p>volume driving lower closing prices is expressed as an index in the usual</p><p>manner.</p><p>Some of you with an eye for detail will notice that if the closing price is</p><p>equal to previous, it is added to both ‘UpDays’ and ‘DownDays’. As the</p><p>closing price has not changed, there is no direction Over yesterday. As</p><p>volume is neither associated with rising or falling price action, it is added to</p><p>both. It seemed an easy method of managing volume on those days where the</p><p>closing price did not change since the day before. Ultimately, I was only</p><p>interested in volume bias that existed within the market.</p><p>The volume bias index defines the relationship between volume that drives</p><p>higher closes and volume that drives lower closes. I considered that if I had</p><p>the ability to define the degree of trading activity relative to the movement of</p><p>the closing price, the level of liquidity would be irrelevant when obvious</p><p>trading activity was not present.</p><p>The volume bias index – levels and percentages</p><p>I will repeat some the previous comments on indicator interpretation as the</p><p>comments here involve volume and not the closing price. It would pay to</p><p>read the following as, although it may appear similar at a glance, it does</p><p>differ slightly from previous comments.</p><p>The index is range bound between 0 and 100 per cent, with arbitrary</p><p>Oversold and Overbought regions at 30 and 70 respectively, as shown in</p><p>table 6.1. This particular indicator is solely reliant on volume for its</p><p>parameters, so highly illiquid stocks will produce inconsistent results under</p><p>certain conditions.</p><p>Table 6.1: values and percentages for volume bias index</p><p>Indicator value Volume % bias</p><p>0 100.0% (bearish)</p><p>10 80.0%</p><p>20 60.0%</p><p>30 40.0%</p><p>40 20.0%</p><p>50 0.0% (neutral)</p><p>60 20.0%</p><p>70 40.0%</p><p>80 60.0%</p><p>90 80.0%</p><p>100 100.0% (bullish)</p><p>The significance of values in the volume bias index is as follows:</p><p>A value of 50 per cent indicates that the total volume associated with</p><p>higher closes is comparable to the total volume associated with lower</p><p>closes.</p><p>A value of 70 per cent reflects a volume bias of 40 per cent in favour of</p><p>higher closing prices compared to the volume driving lower closing</p><p>prices Over the same time period.</p><p>A value of 30 per cent reflects a volume bias of 40 per cent in favour of</p><p>lower closing prices compared to the volume driving higher closing</p><p>prices Over the same time period.</p><p>Stalled price action with an indicator value greater than 70 per cent</p><p>indicates that the buyers have met an equal number of sellers and are no</p><p>longer able to bid prices higher; however, the number of sellers present</p><p>in the market has only managed to stall the uptrend, not reverse it.</p><p>Stalled price action with an indicator value less than 30 per cent</p><p>indicates that the sellers have met an equal number of buyers and are no</p><p>longer able to bid prices lower; however, the number of buyers present</p><p>in the market has only managed to stall the downtrend, not reverse it.</p><p>In addition to understanding the relationship between bullish and bearish</p><p>volume, more specific behaviour can also be put into context, as follows:</p><p>An upward crossOver from the Oversold region indicates that while the</p><p>volume associated with higher closing prices is increasing, the volume</p><p>driving lower closing prices remains dominant Over the same</p><p>period.</p><p>A downward crossOver from the ‘Overbought’ region indicates that</p><p>while the volume associated with lower closing prices is increasing,</p><p>volume driving higher closing prices remains dominant Over the same</p><p>period.</p><p>Uses of the volume bias index</p><p>As the indicator is volume based, its relevance to the analysis process is in</p><p>determining trend support rather than trend maturity — as, for example, the</p><p>relative strength index is used. Further, the volume bias index would not</p><p>always develop in the same manner as other volume tools — such as the</p><p>money flow index, for example.</p><p>It is very common for this particular indicator to lead price action, as it has</p><p>the ability of identifying periods of accumulation and distribution. Often the</p><p>actions of the professionals who attempt to accumulate or distribute a stock</p><p>are busted big time with this method of analysis, simply because volume with</p><p>higher closes is compared with volume on lower closes. Even if the</p><p>professionals want to buy or sell back into the market in order to maintain a</p><p>fairly static share price, the volume bias reflects this. I will give you a fairly</p><p>bad example; however, I think it will convey the concept to which I refer.</p><p>Let’s say that a large professional private trader who manages a private</p><p>portfolio for an oil sheik wants one million shares in an exploration company.</p><p>This level of liquidity in the company simply does not exist, and to place a</p><p>single order on the open market for this many shares would see prices go</p><p>through the roof. The private trader decides to buy, say, 100 000 shares and</p><p>then sell half back into the market. This effectively creates an artificial level</p><p>of supply and demand, which allows the trader to either buy or sell the</p><p>desired number of shares without moving the market excessively. This</p><p>process is known as ramping. Whether such actions are ethical is a separate</p><p>issue; however, it does occur in several forms.1</p><p>Regardless of the method used by any professional, in order to buy shares</p><p>the number of shares purchased must always outweigh the number of shares</p><p>sold — otherwise, the position cannot be filled. Hence, in this situation, the</p><p>volume indicator will still portray a bullish bias in volume activity. Naturally,</p><p>the same applies to the distribution of a stock nearing rally conclusion. For a</p><p>large position to be closed the number of shares sold must always outnumber</p><p>the number shares purchased, meaning a bearish bias in volume activity will</p><p>be identified.</p><p>It would also be logical to expect that in buying conditions the closing price</p><p>will be higher, as the astute professional would only sell back into the market</p><p>as required in an attempt to hold prices at current levels. This would only</p><p>occur as prices begin to rise excessively. Volume with any selling will still be</p><p>consistently less than buying volume — otherwise, the trader is effectively</p><p>lightening the position, which is not the aim.</p><p>Figure 6.1 shows a chart of EquiGold with the volume bias index at the</p><p>bottom.</p><p>Figure 6.1:- volume bias index</p><p>The indicator at the bottom of figure 6.1 defines the volume bias between</p><p>bullish and bearish activity. From February 2005, EquiGold was experiencing</p><p>a strong downtrend. Conventional analysis would lead most people to believe</p><p>that, as this is a downtrend on heavy volume, the bears are having a field day</p><p>as pessimism reigns supreme. Unfortunately, this analysis is again drawn at</p><p>face value. The volume bias index shows a completely different view of the</p><p>same decline. As prices declined, volume increased when EQI closed higher.</p><p>This is highlighted by the rise in the volume bias index. As volume with</p><p>higher closes is greater than volume associated with lower closes, it can</p><p>reasonably be assumed that increased accumulation is occurring as prices</p><p>decline.</p><p>By the time the downtrend concludes, volume bias has become positive,</p><p>with an indicator value greater than 50 per cent. Bullish volume now</p><p>dominates EQI. How valuable do you think this type of information can be?</p><p>This particular downtrend effectively concluded on buying pressure, not</p><p>seller exhaustion and, as you know, if a trend concludes on buying pressure,</p><p>probability strongly favours an ensuing rally.</p><p>Volume bias channel</p><p>I’m sure some of you will have seen where I was going with this. Have you</p><p>ever thought about combining price and volume on one chart? I feel</p><p>reasonably confident that the majority of traders will have given little or no</p><p>consideration to this particular line of analysis. The volume bias index is</p><p>relative to market sentiment and is also relevant to ongoing trend</p><p>development.</p><p>Figure 6.2 shows the chart of EQI with the volume bias channel applied.</p><p>Figure 6.2: volume bias channel</p><p>To create the channel shown in figure 6.2, I have continued with the half-</p><p>lunar concept in terms of indicator periods. As the chart shows, setting the</p><p>channel at 14 periods is nothing short of a complete disaster. There is no</p><p>correlation between general trend development and volume activity. Sure, the</p><p>indicator at the bottom of the screen looks nice, but it does little to assist in</p><p>the definition of price action relative to the volume bias that is underpinning</p><p>trend development. While the peaks and troughs in indicator behaviour may</p><p>coincide with the highs and lows in trend development, the real picture</p><p>remains hidden. The channel still does not identify the points of dominant</p><p>bullish behaviour.</p><p>Establishing a benchmark</p><p>Who’s up for really pushing the envelope? Would it be possible to trade</p><p>successfully using a price based channel derived purely from volume?</p><p>Remember — the volume bias channel defines nothing more than a volume</p><p>bias Over a specified period. If volume is heavier on higher closing prices,</p><p>the indicator will rise; however, the indicator cannot increase in unison with</p><p>the trend if volume is lacking.</p><p>Testing parameters</p><p>The testing parameters used to create the benchmark are as follows:</p><p>channel periods — 14</p><p>entry trigger — a close above the lower cord of the bullish band</p><p>exit trigger — a close below the upper cord of the bearish band or a</p><p>close below the upper cord of the bullish band for price action that has</p><p>moved into the Overbought region.</p><p>Test results</p><p>The test results for the volume bias channel benchmark, also shown in table</p><p>6.1, are as follows:</p><p>The first problem is the sheer number of potential signals that have</p><p>been identified. With 4024 potential trades passing Over the screen, you</p><p>would be deluding yourself if you genuinely believed you had the ability</p><p>to process this amount of information successfully in real time.</p><p>The peak to valley drawdown was disappointing — to the point of</p><p>becoming unacceptable. While you should not get too carried away with</p><p>the worst-case scenario, 18.13 per cent is not ideal by any stretch of the</p><p>imagination.</p><p>The ratio of successful trades is 35.29 per cent, which, although this is</p><p>an improvement Over previous testing results produced by more accepted</p><p>techniques, is unacceptable for real time application.</p><p>The average loss is satisfactory at $281.96, while the worst trade during</p><p>the testing period was -$1,647.14. While most people would rather not</p><p>experience a loss of this magnitude, providing losses of this size were not</p><p>too frequent, they would not be wiped out by this amount.</p><p>The annualised rate of return is more than acceptable, coming in at</p><p>24.99 per cent.</p><p>Table 6.2: volume bias channel — benchmark</p><p>Profit summary</p><p>Profit status PROFITABLE</p><p>Starting capital $25 000.00</p><p>Finishing capital $89 850.96</p><p>Maximum equity/(date) $64 850.96 (30/09/2005)</p><p>Minimum equity/(date) −$2627.68 (14/07/2000)</p><p>Gross trade profit $120 679.49 (482.72%)</p><p>Gross trade loss −$55 828.53 (−223.31%)</p><p>Total net profit $64 850.96 (259.40%)</p><p>Average profit per trade $211.93</p><p>Profit factor 2.1616</p><p>Profit index 53.74%</p><p>Total transaction cost $12 240.00</p><p>Total slippage $0.00</p><p>Daily compound interest rate 0.0611%</p><p>Annualised compound interest rate 24.9938%</p><p>Trades processed 4024</p><p>Trades taken 306</p><p>Partial trades taken 0</p><p>Trades rejected 3718</p><p>Winning trades 108</p><p>(35.29%)</p><p>Losing trades 198 (64.71%)</p><p>Largest winning trade/(date) $26 209.58 (11/08/2000)</p><p>Largest losing trade/(date) −$1647.14 (30/06/2000)</p><p>Average winning trade $1117.40</p><p>Average losing trade −$281.96</p><p>Average win/average loss 3.9630</p><p>Trade duration statistics (all trades)</p><p>Maximum trade duration 400 (days)</p><p>Minimum trade duration 6 (days)</p><p>Average trade duration 55 (days)</p><p>Winning trades</p><p>Maximum trade duration 400 (days)</p><p>Minimum trade duration 7 (days)</p><p>Average trade duration 105 (days)</p><p>Losing trades</p><p>Maximum trade duration 182 (days)</p><p>Minimum trade duration 6 (days)</p><p>Average trade duration 29 (days)</p><p>Consecutive trade statistics</p><p>Maximum consecutive winning trades 6</p><p>Maximum consecutive losing trades 10</p><p>Average consecutive winning trades 1.59</p><p>Average consecutive losing trades 2.91</p><p>Relative drawdown</p><p>Maximum dollar drawdown/(date) $5618.61 (6/10/2000)</p><p>Maximum percentage drawdown/(date) 12.9900% (14/07/2000)</p><p>Absolute (peak to valley) dollar drawdown</p><p>Maximum dollar drawdown $9366.88 (18.1300%)</p><p>Capital peak/(date) $51 675.26 (11/08/2000)</p><p>Capital valley/(date) $42 308.38 (23/02/2001)</p><p>What is interesting so far is that ‘flash’ and quite often relatively complex</p><p>methods of evaluating price action have been unable to match volume with</p><p>regard to consistency. This would tend to confirm that the real information is</p><p>concealed within price and volume, and that those who take the time to</p><p>analyse price action and to understand the underlying relationship between</p><p>price and volume will usually perform with greater consistency than those</p><p>who seek a more complex approach.</p><p>Equity curve and yearly profits</p><p>A look at the closed equity curve in figure 6.3 shows that once the bull run</p><p>commenced in March 2003, the general trend was upward — albeit a bit on</p><p>the unstable side occasionally. The returns prior to the bull run were</p><p>disappointing, with no major headway in the growth of trading capital in a</p><p>two-year period.</p><p>Figure 6.3: closed equity curve for 14-period volume bias channel</p><p>Figure 6.4: shows the yearly profits for the volume bias channel.</p><p>Figure 6.4: yearly profits for 14-period volume bias channel</p><p>As shown in figure 6.4, 2002 produced the only negative result with a</p><p>$1813.62 loss. This is acceptable when the fact that the market declined by</p><p>12.25 per cent during this period is taken into account. Looking at the equity</p><p>curve with yearly profits, the biggest concern is that 2000 produced a sudden</p><p>profit jump in September, indicating that the profit in 2000 came from a</p><p>limited number of trades or possibly a single trade. This is not a desirable</p><p>outcome for any strategy. While I have never been a believer in the straight</p><p>line theory with regard to equity curves — as this tends to suggest that the</p><p>strategy cuts losses but fails to let profits run — an equity curve that deviates</p><p>in this manner is also unacceptable. It is nowhere near as bad, however, as</p><p>some of the other equity curves produced by highly respected confirmation</p><p>tools.</p><p>The end of the benchmarks</p><p>In this and the previous chapters, I have developed benchmarks to use for</p><p>further comparison as I attempt to either confirm or disprove that relative</p><p>volume analysis is actually relative to ongoing trend development.</p><p>Some of you may be sick of these testing results; however, if your intention</p><p>is to not become cannon fodder for the seasoned trader, you need to know</p><p>what you are up against. Believe me, the old ostrich trick of inserting your</p><p>head in the sand, or saying something doesn’t apply to you because you’re a</p><p>part-timer, just doesn’t cut the mustard. Part-timers, in all probability, are at</p><p>greater risk simply because they are more likely to lack the innate feel that</p><p>full-time traders have, no matter how they may choose to delude themselves.</p><p>The market has no respect for anyone’s delicate ego, nor does it offer a</p><p>refund for inexperience or concessions for stupidity and ignorance.</p><p>In the following chapters, I move beyond the benchmarks and begin testing</p><p>the concepts for compatibility, confirmation and bilateral relationships.</p><p>Coding for chapter 6</p><p>Volume bias index — BullCharts</p><p>[Target=Percent; author=Book, Leon]</p><p>Pds:=input("Volume Periods", 34, 1, 250);</p><p>Smoothing:=input("Smoothing", 1, 1, 55);</p><p>Trigger:=input("Trigger Line", 1, 1, 55);</p><p>Value2:=input("Over Bought", 70, 51, 99);</p><p>Value3:=input("Over Sold", 30, 1, 49);</p><p>Value4:=input("Upper Neutral Zone", 55, 50, 99);</p><p>Value5:=input("Lower Neutral Zone", 45, 1, 50);</p><p>Up:=(Sum(If(Close>=Ref(Close,-1),Volume,0),pds))/pds;</p><p>Down:=(Sum(If(Ref(Close,-1)>=Close,Volume,0),pds))/pds;</p><p>Bias:=100-(100/(1+(Up/Down)));</p><p>Index:= Ma(Bias,Smoothing,E);</p><p>Trigg:=Ma(Ma(Bias, Smoothing, E), Trigger, E);</p><p>[Color=Black;Linestyle=Solid]</p><p>Index;</p><p>[Color=Blue;Linestyle=dash]</p><p>Trigg;</p><p>[Color=Light Slate Gray;Linestyle=Dotted]</p><p>value3; value2;</p><p>[name=1st Zone1]</p><p>Value5;</p><p>[name=2nd Zone2]</p><p>value4;</p><p>[name=Bullish Zone; linestyle=Fill;color=light Slate Gray]</p><p>Value2; Value4;</p><p>[name=Neutral Zone; linestyle=Fill;color=White]</p><p>value5; Value4;</p><p>[name=Bullish Zone; linestyle=Fill;color=Dark Grey]</p><p>Value3; Value5;</p><p>{Copyright Leon Wilson, 2005}</p><p>Volume bias channel — BullCharts</p><p>[Target=Price; author=Book, Leon]</p><p>Pds:=input("Volume Periods", 34, 1, 250);</p><p>Smoothing:=input("Smoothing", 1, 1, 55);</p><p>Value2:=input("Over Bought", 70, 50, 99);</p><p>Value3:=input("Over Sold", 30, 1, 50);</p><p>Value4:=input("Upper Neutral Zone", 55, 50, 99);</p><p>Value5:=input("Lower Neutral Zone", 45, 1, 50);</p><p>Up:=(Sum(If(Close>=Ref(Close,-1),Volume,0),Pds))/Pds;</p><p>Down:=(Sum(If(Ref(Close,-1)>=Close,Volume,0),Pds))/Pds;</p><p>Index:= 100-(100/(1+(Up/Down)))</p><p>[Color=Light Slate Gray;Linestyle=Dotted]</p><p>OB:=Ma((Index -Value3),Smoothing,E);</p><p>OS:=Ma((Index -Value2),Smoothing,E);</p><p>NZU:=Ma((Index -Value4),Smoothing,E);</p><p>NZL:=Ma((Index -Value5),Smoothing,E);</p><p>{Show Results}</p><p>[name=OverB;]</p><p>Close-(Close*(OB/100));</p><p>[name=OverS;]</p><p>Close-(Close*(OS/100));</p><p>[name=NeutUp;]</p><p>Close-(Close*(NZU/100));</p><p>[name=NeutLower;]</p><p>Close-(Close*(NZL/100));</p><p>{Upper Channel}</p><p>[name=Upper; linestyle=Fill;color=Light Slate Gray]</p><p>Close-(Close*(NZU/100)); Close-(Close*(OS/100));</p><p>{Neutral Zone}</p><p>[name=Zone; linestyle=Fill;color=White]</p><p>Close-(Close*(NZU/100)); Close-(Close*(NZL/100));</p><p>{Lower Channel}</p><p>[name=Lower ; linestyle=Fill;color=Dark Grey]</p><p>Close-(Close*(NZL/100));Close-(Close*(OB/100));</p><p>{Copyright Leon Wilson, 2005}</p><p>Volume bias index — MetaStock</p><p>Pds:=Input("Volume Periods", 1, 250, 34);</p><p>Smoothing:=Input("Smoothing", 1, 55, 1);</p><p>Value2:=Input("Over Bought", 50, 99, 70);</p><p>Value3:=Input("Over Sold", 1, 50, 30);</p><p>Value4:=Input("Upper Neutral Zone", 50, 99, 55);</p><p>Value5:=Input("Lower Neutral Zone", 1, 50, 45);</p><p>Up:=(Sum(If(CLOSE>=Ref(CLOSE,-1),VOLUME,0),pds))/pds;</p><p>Down:=(Sum(If(Ref(CLOSE, -1)>=CLOSE,VOLUME,0),pds))/pds;</p><p>Bias:=100-(100/(1+(Up/Down)));</p><p>Index:= Mov(Bias,Smoothing,E);</p><p>Trigg:=Mov(Mov(Bias,Smoothing,E),1,E);</p><p>Index; Trigg; Value3; Value2; Value5; Value4;</p><p>{Copyright Leon Wilson, 2005}</p><p>Volume bias channel — MetaStock</p><p>Pds:=Input("Volume Periods", 1, 250, 34);</p><p>Smoothing:=Input("Smoothing", 1, 55, 1);</p><p>Value2:=Input("Over Bought", 50, 99, 70);</p><p>Value3:=Input("Over Sold", 1, 50, 30);</p><p>Value4:=Input("Upper Neutral Zone", 50, 99, 55);</p><p>Value5:=Input("Lower Neutral Zone", 1, 50, 45);</p><p>Up:=(Sum(If(CLOSE>=Ref(CLOSE,-1),VOLUME,0),Pds))/Pds;</p><p>Down:=(Sum(If(Ref(CLOSE,-1)>=CLOSE,VOLUME,0),Pds))/Pds;</p><p>Index:= 100-(100/(1+(Up/Down)));</p><p>OB:=Mov((Index -Value3),Smoothing,E);</p><p>OS:=Mov((Index -Value2),Smoothing,E);</p><p>NZU:=Mov((Index -Value4),Smoothing,E);</p><p>NZL:=Mov((Index -Value5),Smoothing, E);</p><p>CLOSE-(CLOSE*(OB/100));</p><p>CLOSE-(CLOSE*(OS/100));</p><p>CLOSE-(CLOSE*(NZU/100));</p><p>CLOSE-(CLOSE*(NZL/100));</p><p>{Copyright Leon Wilson, 2005}</p><p>Volume bias channel — TradeSim (MetaStock)</p><p>Pds:=34;</p><p>Up:=(Sum(If(CLOSE>=Ref(CLOSE,-1),VOLUME,0),Pds))/Pds;</p><p>Down:=(Sum(If(Ref(CLOSE,-1)>=CLOSE,VOLUME,0),Pds))/Pds;</p><p>Index:= 100-(100/(1+(Up/Down)));</p><p>NZU:=Index -55;</p><p>NZL:=Index -45;</p><p>UC:= Index -85;</p><p>NeutUp:=CLOSE-(CLOSE*(NZU/100));</p><p>OB:=CLOSE-(CLOSE*(UC/100));</p><p>NeutL:= CLOSE-(CLOSE*(NZL/100));</p><p>EntryTrigger := CLOSE>NeutUp AND (Ref(CLOSE,-1)<=Ref(NeutUp,-1));</p><p>EntryPrice := OPEN;</p><p>ExitTrigger := CLOSE<NeutL AND (Ref(CLOSE,-1) >=Ref(NeutL,-1)) OR CLOSE<OB</p><p>AND (Ref(CLOSE,-1) >=Ref(OB,-1));</p><p>ExitPrice := OPEN;</p><p>InitialStop:=0;</p><p>{Copyright Leon Wilson, 2005}</p><p>1 While I cannot confirm it personally, I have read several articles</p><p>suggesting that ramping is common on some international markets,</p><p>especially where cheap stocks are concerned. Larger blue chips are under</p><p>the eye of the authorities and with a much higher trading price they are less</p><p>likely to experience price manipulation, if for no other reason than the</p><p>significant amount of capital required to move the market.</p><p>I received an email some time ago from an individual who was concerned</p><p>that a stock was being driven upward in price only to be dumped. This is</p><p>commonly known as ‘pump and dump’, a scheme that is most likely to</p><p>occur in very low price stocks. Following a couple of heavy hits on the</p><p>trading capital, a desperate plea for assistance was received. How can</p><p>people trade a stock that they think is deliberately being pumped? My</p><p>answer was simple — don’t! Naturally, the idea is to catch the rise in price;</p><p>however, you can never pick the dumping point because normal market</p><p>forces are not at work. This is artificial behaviour, and any suspect trading</p><p>activity should always be avoided.</p><p>Chapter 7</p><p>Compatibility</p><p>The single biggest mistake that any trader can make is to simply take</p><p>accepted practice at face value. By adapting various indices to price action,</p><p>the compatibilities of existing confirmation tools and the parameters applied</p><p>can be gauged.</p><p>The benchmarks tested in the previous chapters have shown that using</p><p>channels set at 14 periods is out of whack with trend development. In this</p><p>chapter, I will move beyond the benchmark by changing the parameters —</p><p>that is, the periods used, and the entry and exit signals.</p><p>Tip</p><p>Never apply an indicator at specific values simply because it appears to be compatible on</p><p>the surface. The fact that a parameter is popular does not necessarily imply that it is</p><p>practical or beneficial. The indicator you use must be relative to trend development and</p><p>your strategy style — otherwise, it will be of no benefit in real time.</p><p>Relative relationships – 14 periods</p><p>It is often surprising just how poorly default parameters actually relate to</p><p>general trend development. When I initially devised this particular technique</p><p>and finally managed to adapt the few of the more respectable indicators to</p><p>price action, I was often disappointed with what appeared before me. Often</p><p>the default value for the indicators included with most technical trading</p><p>programs is 14 periods, which looks more than acceptable when applied in a</p><p>classic manner at the bottom of the chart. However, applying the indicator</p><p>directly to price action created a less than impressive relationship, especially</p><p>for position traders.</p><p>The half-lunar cycle?</p><p>The majority of inexperienced traders will at some point apply indicators</p><p>because they read about them in magazine articles — or in the glossy sales</p><p>brochure that promotes the latest super secret exclusive ‘Cosine Reduced</p><p>Adaptive Percentage’ index — or see the process being applied in</p><p>workshops. The indicator will be applied at the usual default of 14 periods —</p><p>because just about every range-bound index is applied at 14 periods — while</p><p>it is accompanied by a ten-line paragraph on how to interpret it. The</p><p>inexperienced trader then unwittingly applies this abomination, usually</p><p>unsuccessfully, to price action. Like everyone else, I was guilty of this</p><p>behaviour early in my career simply because I knew no different and trading</p><p>software was relatively new at the time; however, unlike the majority, I</p><p>refused to accept this relationship as sound at face value simply because I</p><p>was told, ‘Trust me, it works a treat at 14 periods’. The alarm bells started to</p><p>ring when I heard that this value, the most popular of indicator values, was</p><p>based on a half-lunar cycle. This seemed odd as people seldom think in this</p><p>manner. The thought pattern for the majority is usually along the following</p><p>time frames:</p><p>Weekly — Over the short term, most people think along a weekly time</p><p>frame, such as ‘I must service the car this week’ or ‘I’m having this</p><p>weekend off’. I am yet to hear anyone say that the car will be serviced</p><p>next lunar cycle.</p><p>Monthly — the second time frame is usually monthly. Most people will</p><p>plan to accomplish certain goals each month — whether it’s dodging the</p><p>gardening by going on a well-planned fishing trip, such as in my case, or</p><p>‘I’m having a month off to put my feet up’. The average number of</p><p>trading days per month is around 21, which would help explain why a</p><p>21-day moving average works well for short-term analysis. A lunar</p><p>cycle is 28 days, so still no relationship there.</p><p>Six months — holidays and larger projects such as selling the house</p><p>often involve a longer time frame. In trading days this is reasonably</p><p>close to 125 days, so I can see why a 125-day moving average tends to</p><p>test with a degree of consistency.</p><p>Yearly (calendar) — the majority of people review their longer term</p><p>outlook yearly, which would also explain why 52-week highs tend to</p><p>produce consistent results.</p><p>Please understand that I am not a psychologist; however, it seems the human</p><p>psyche generally works on cycles of weeks, months and years. As everyone</p><p>is directly influenced by the same economic forces, it would be logical to</p><p>analyse the market from this perspective. While there may have been sound</p><p>reasoning at the time for 14 periods, in-depth testing since tends to confirm</p><p>that there is no correlation between 14-period indicators and superior results.</p><p>I needed to see how the indicator related to actual trend development and,</p><p>as the previous chapters show, the picture I was starting to receive was not a</p><p>pretty one. Beyond the extreme points in indicator value there is no relevance</p><p>to trend development, which is what an indicator is supposed to portray. The</p><p>majority of indicators focus on the closing price with the intention of putting</p><p>the behaviour of the closing price in context with trend development. How on</p><p>earth can an indicator put the closing price into context with trend</p><p>development if its underlying value is not relative to trend development? In</p><p>reality, all you are doing is following the behaviour of the closing price using</p><p>a random value hoping that at some point it will mesh with trend</p><p>development. The problem is potentially twofold when popular indicators are</p><p>used:</p><p>The indicator is useless and makes for a nice marketing gimmick.</p><p>The indicator is not in sync with trend development, meaning the</p><p>indicator is primarily irrelevant to price action as it unfolds.</p><p>Figure 7.1 shows the relative price channel set at the default 14 periods.</p><p>Figure 7.1: default 14-period relative price channel</p><p>The chart shown in figure 7.1 needs little explanation. It is a real mess</p><p>throughout 2004 with little or no correlation existing between price action</p><p>and indicator development. Further, the highs and lows in price action of</p><p>2005 have absolutely no correlation with the peaks and troughs in channel</p><p>development.</p><p>It would appear that it doesn’t really matter what the underlying trend may</p><p>be doing, there is a distinct lack of continuity between the much-loved RSI</p><p>and price action at 14 periods.</p><p>Perhaps you can now begin to understand why seasoned traders</p><p>consistently argue that inexperienced traders should forget about chasing the</p><p>perfect indicator, the trading version of the Holy Grail, and take the time to</p><p>understand price action. These seasoned traders may not be able to pinpoint</p><p>the reason behind so many indicators underperforming in real time, but they</p><p>have learnt Over time and from repeated experience that to rely on an</p><p>indicator without serious assessment first is usually fraught with danger. The</p><p>inexperienced trader runs from one indicator to the next, looking for that</p><p>magical solution, only to quickly dismiss the latest choice as unacceptable</p><p>before moving on. The reason the inexperienced trader can</p><p>never find an</p><p>indicator that truly helps is because the indicator is not in sync with price</p><p>action or trend development. It will not matter how many indicators a novice</p><p>trader pulls from the menu — if their parameters are incompatible with the</p><p>underlying trend, the indicators are never going to produce the results the</p><p>trader desires, regardless of how brilliant the concept may be. The trader may</p><p>have already had the ideal indicator in his or her grasp, but incompatible</p><p>application brought about by blindly following accepted practice without</p><p>question meant that only another ordinary result was achieved. If you want to</p><p>avoid becoming entangled in incompatible practice when position trading,</p><p>your perspective must be relative to trend development.</p><p>Defining correct periods</p><p>Many traders will reduce their indicator periods to 10 or less, while others</p><p>will expand their time frame to 21 periods or longer. The fewer the number of</p><p>periods applied, the more volatile the readings produced by the indicator will</p><p>be. This is because a smaller change in daily activity is required to</p><p>significantly shift the indicator value, due to the reduced number of</p><p>calculation periods being applied. Alternatively, indicators that are calculated</p><p>Over an extended period tend to produce more consistent values and are less</p><p>likely to generate those exceptionally high values usually experienced by</p><p>their shorter time period counterparts. A disappointing aspect of the kind of</p><p>indicator application undertaken by the majority of traders is the lack of</p><p>research that underpins the values chosen. The bulk of indicator parameters</p><p>are chosen not through exhaustive testing and research but for the sole</p><p>purpose of adding a pinch of individuality, under the illusion that the impact</p><p>of an alternative value will be minimal at best. If you apply this approach,</p><p>you obviously have no idea just how wrong and dangerous this practice can</p><p>be.</p><p>Why do so many range-bound indices fail to deliver? I have often heard</p><p>statements along the lines of, ‘I confirm a bullish or bearish break-out with a</p><p>turning RSI . . . blah, blah, blah’. You know the drill; all very textbook. I</p><p>have seen this method of analysis repeatedly flogged as a sound strategy. In</p><p>reality, however, if price action continues, it is usually more a coincidence</p><p>than genuine confirmation. I can hear the feet stamping from here — cries of</p><p>denial and ‘Wilson’s an idiot’ — but think about it. The RSI displays the</p><p>average bias that exists between higher closes and lower closes. What it does</p><p>not take into account is the total number of days associated with higher and</p><p>lower closes beyond the averaging process.</p><p>Remember how range-bound indices are calculated. While there are many</p><p>coding variations for the relative strength index, the following is one of the</p><p>more common:</p><p>100-(100/(1+(UpDay/DownDay)))</p><p>The following is a basic example that aims to highlight the calculation</p><p>process as each day unfolds. While trading programs use exponential moving</p><p>averages, these are far too complex for manual calculation, so for this</p><p>example I have used a simple averaging process for defining the appropriate</p><p>values as needed. Try to focus on the application rather than specific values.</p><p>In the example, table 7.1 represents the bias experienced by the closing</p><p>price Over 14 periods for a stock (Consistent values have been applied for</p><p>ease of calculation.) For example, on day six the closing price finished $0.50</p><p>higher than day seven, and day nine closed $0.20 lower than day ten.</p><p>Remember that the RSI and similar range-bound indices do not focus on the</p><p>value of the closing price but rather the change in value experienced by the</p><p>closing price. Over the 14 days, there were 11 days where the closing price</p><p>was lower (shown in bold) and three days where the closing price finished</p><p>higher than previous (shown in italics).</p><p>Table 7.1: bias experienced by the closing price — yesterday ($)</p><p>Using the above data, the ‘UpDay’ is determined as follows:</p><p>UpDay = ($0.50 + $0.50 + $0.50) /14</p><p>= $0.107.</p><p>The ‘DownDay’ is determined as:</p><p>DownDay = ($0.20 + $0.20 + $0.20 + $0.20 + $0.20 + $0.20 + $0.20 + $0.20</p><p>+ $0.20 + $0.20 +$0.20)/14</p><p>= $0.157.</p><p>This means the RSI is determined as:</p><p>RSI =100-(100/(1+(0.107/0.157)))</p><p>= 100-(100/(1+0.6818))</p><p>= 100-(100/1.6818)</p><p>= 100-(59.45946)</p><p>= 40.54054 per cent.</p><p>Table 7.2 shows the data for the example share on the day after the data</p><p>shown in table 7.1, or ‘today’. As the majority of indicators produce a rolling</p><p>value — that is, they step forward as new data is loaded — you will notice</p><p>that the values have all stepped one day to the left. Day six has now become</p><p>day seven, and so on, with the latest day’s data being added to the front of the</p><p>list as day one. In this example, today has closed $0.20 higher than yesterday.</p><p>Table 7.2: bias experienced by the closing price — today ($)</p><p>Hence, the ‘UpDay’ is now:</p><p>UpDay = ($0.50 + $0.50 + $0.50 + $0.20) /14</p><p>= $0.121.</p><p>The ‘DownDay’ is determined as:</p><p>DownDay = ($0.20 + $0.20 + $0.20 + $0.20 + $0.20 + $0.20 + $0.20 +</p><p>$0.20 + $0.20 + $0.20)/14</p><p>= $0.157.</p><p>The RSI is now:</p><p>RSI =100-(100/(1+(0.121/0.143)))</p><p>= 100-(100/(1+0.846))</p><p>= 100-(100/1.846)</p><p>= 100-(54.05405)</p><p>= 45.94595 per cent.</p><p>What observations can you make about the above? Today’s closing price has</p><p>finished $0.20 higher than yesterday, meaning it has done nothing more than</p><p>offset the lower closing price experienced by day two. Even though day one</p><p>achieved nothing more than the negation of the previous day’s close, the RSI</p><p>increased by 13.3 per cent. Taking this further, even though 10 of the 14 days</p><p>have consistently closed lower, the RSI ticked up strongly. While the</p><p>dominance of lower closes is reflected by an RSI value that is less than 50 per</p><p>cent, it fails to highlight the number of lower closes Over the calculation</p><p>period. A rising RSI indicates little about the trend in place and confirms</p><p>nothing, given the nature of the calculation process. Just one or two bullish or</p><p>bearish periods have the ability to swing the RSI significantly and</p><p>disproportionately influence the averaging process. This also explains why</p><p>the RSI and similar range-bound indices tend to mirror price action.</p><p>An upward move of the RSI is not confirmation</p><p>By design, the RSI must swing upward on the back of a higher closing price.</p><p>It doesn’t matter how significant or inconsequential the higher closing price</p><p>is, the upward range is added to the bullish element of the calculation process</p><p>while bearish activity scores a zero. The determining factor that influences</p><p>the degree of the change experienced by the RSI is the number of periods</p><p>used. How can any indicator that, by design, must follow a higher closing</p><p>price be considered an independent confirmation tool? In short, it cannot be.</p><p>Even though the above example is fairly basic, what becomes readily</p><p>apparent is that the inclusion of a single value to either the bullish or bearish</p><p>element of the calculation process must affect the direction of the index. To</p><p>use a rising RSI or similar to confirm a solid rise in price action accomplishes</p><p>nothing. Some will suggest that a rising RSI that is accompanied by</p><p>resistance break-outs or bullish trendline penetration is a strong signal. The</p><p>strong signal is the break of resistance or the trendline penetration — it has</p><p>nothing to do with a rising RSI. By design, the RSI must follow price action</p><p>as it breaks resistance or pushes upward through the trendline. To suggest</p><p>that this is confirmation is incorrect at the most fundamental level of</p><p>confirmation analysis.</p><p>Using some lateral thinking, it would seem reasonable to conclude that if</p><p>the RSI mirrors price action in a similar manner as a short-term moving</p><p>average, both tools should be treated similarly. For example, very few traders</p><p>who focus on position trading will apply a seven-day moving average, if for</p><p>no other reason than it is irrelevant to the underlying strategy. The moving</p><p>average is simply too short in duration and accomplishes nothing more than</p><p>replicating price action and</p><p>so producing an excessive number of false</p><p>signals. Position traders slow the moving average down, usually to</p><p>somewhere between 21 and 34 periods, in order to remove much of the</p><p>market noise. While the moving average is less responsive, this is more than</p><p>compensated by the continuity in development and improved consistency.</p><p>While the RSI is not a moving average, its characteristics are not that</p><p>dissimilar. If slowing a moving average down improves reliability while</p><p>removing excessive market noise, it would seem reasonable to conclude that</p><p>the same principles can be applied to the RSI. While accepted practice leads</p><p>traders to believe that short-term confirmation indicators lead to a more</p><p>responsive affirmation of the entry trigger, my research tends to suggest that</p><p>the opposite is true. While your natural action may be to shorten the number</p><p>of confirmation periods in order to improve results, generally the opposite</p><p>action is required. This is supported by the results in this book, where all</p><p>slower relative price channels produced a more consistent result.</p><p>Making the indicators relative</p><p>If your intention was to position trade using the RSI as a confirmation tool</p><p>then adapting the existing indicators to price action will allow you to identify</p><p>the parameters that are relative to general trend development. Wasn’t there an</p><p>old scientist somewhere who came up with the theory of relativity? Definitely</p><p>a trader at some point I would say. Unlike the scientific equivalent, the</p><p>process of relativity analysis is not complex. It’s simply a matter of defining</p><p>a set of parameters that more accurately reflect general trend development.</p><p>The process of defining the relationship between the relative price channel</p><p>and price action starts by analysing the relationship visually. Once the peaks,</p><p>troughs and trends have been synchronised visually, you can finetune the</p><p>relationship for improved efficiency by testing crossOver points.</p><p>Figure 7.2 shows the directional movement channel at 34 periods.</p><p>Figure 7.2: directional movement channel at 34 periods</p><p>Looking at the chart shown in figure 7.2, I settled on 34 periods as a suitable</p><p>starting point from a visual perspective.</p><p>The reasons are as follows:</p><p>Throughout 2002 EQI experienced a bullish consolidation as it</p><p>developed in the bullish band of the directional movement channel.</p><p>The bullish rally is fairly self-explanatory — trending price action again</p><p>develops in the bullish band while consistently testing the upper cord.</p><p>This is strong bullish activity.</p><p>The neutral rally that develops throughout the bulk of 2004 will be seen</p><p>by some as an opportunity to enter; however, this particular rally can be</p><p>seen to be developing on relatively neutral activity with no prominent</p><p>bullish support underpinning trend development. It is unlikely that a</p><p>rally of this nature will survive long term unless the underlying support</p><p>improves dramatically.</p><p>The bearish decline commencing in late 2004 is exactly that — a strong</p><p>bearish rally consistently testing the lower cord of the bearish band.</p><p>Finally the rally of May 2005 is a bullish rally in a bearish climate. We</p><p>know this because the bullish trend remains clearly defined within the</p><p>upper and lower cords of the bearish band. Perhaps you have heard the</p><p>statement, ‘Rally in a downtrend’, but have never been able to clearly</p><p>identify its appearance due to the degree of subjectivity. Conventional</p><p>charts show that there is bullish price action combined with a bearish</p><p>indicator, so this is commonly referred to as a ‘non-confirmation’. This</p><p>is a polite way of saying, ‘I’m not sure about this one’. The optimist will</p><p>see rising price action that is accompanied with rising indicator values</p><p>and so believe it is a rally on rising strength. This is all the evidence</p><p>required in order to initiate an entry. Indeed, this type of set-up is often</p><p>sought by the more aggressive traders among us.</p><p>However, through looking at the channel, it is clear that this is a bullish rally</p><p>in bearish conditions. As a bullish rally in bearish conditions can be</p><p>confirmed, the logic behind considering a long side position must be</p><p>questioned.</p><p>As can be seen in figure 7.2, it seems using 34 periods means the channel</p><p>will more accurately reflect general trend development. This will be tested</p><p>later in this chapter.</p><p>Stand-alone confirmation</p><p>In order to ascertain whether the relative adaptation of conventional</p><p>indicators to price action is genuinely effective as a means of legitimate and</p><p>effective confirmation, I will continue to use the indicator as a stand-alone</p><p>tool, as per the benchmarks. However, I will move beyond the benchmarks</p><p>by adjusting the entry and exit signals.</p><p>So far I have applied accepted practice; however, in order for the channels</p><p>to be applied from a perspective of relativity, the entry points also need to</p><p>reflect a degree of practical application. Hence, the entry signal is as follows:</p><p>A position is opened on the first close that occurs above the lower cord</p><p>of the bullish band.</p><p>A position is taken on the first signs of bullish activity as indicated by the</p><p>first closing price that occurs within the bullish band. The aim is to enter</p><p>relatively early in trend development while also ensuring that any potentially</p><p>nasty rebounds from bearish conditions are avoided.</p><p>When trading, it is important to lock in profits from runaway price action.</p><p>This is identified by a close below the upper cord of the bullish band. The</p><p>other critical element associated with being a position trader is absorbing</p><p>retracement behaviour and periods of inactivity. I have found that</p><p>consolidation periods often develop within the neutral zone, so it would be</p><p>logical to exit when the closing price dropped below the upper cord of the</p><p>bearish band.</p><p>Hence, a position is closed on either:</p><p>a close below the upper cord of the bullish band</p><p>a close below the upper cord of the bearish channel.</p><p>Some may consider that I have made a typing error by stipulating that the</p><p>close must occur below the upper cord of the bearish channel; however, I</p><p>have consistently found that pauses in trending activity will often see price</p><p>action drift into the neutral zone before continuing. Extensive testing and</p><p>research Over an extended period consistently produced superior results</p><p>when the exit is based on a close below the upper cord of the bearish band</p><p>rather than a close below the lower cord of the bullish band. Remember —</p><p>conventional thinking leads you to believe that a value at or immediately</p><p>below 50 per cent with a range-bound index is bad, which is absolute</p><p>nonsense. It merely tells you that both bullish and bearish activity is</p><p>comparable Over the same period. Only when an index falls some distance</p><p>below 50 per cent do you have a genuine bearish influence Over price action.</p><p>The identification of neutral activity needs to be a range not a precise value.</p><p>The neutral range I apply, after exhaustive research, is between the values of</p><p>45 per cent and 55 per cent. In the majority of cases, this range defines the</p><p>bulk of unbiased price action or marginal rallies with a fair degree of</p><p>consistency, which is the underlying essence of trading — consistency</p><p>brought about by the realm of probability.</p><p>The entry and exit signals outlined above will be used for all strategies</p><p>tested in this chapter.</p><p>Figure 7.3 shows the price action of GNS, highlighting where entry and</p><p>exit signals would occur using the above strategy.</p><p>Figure 7.3: directional movement channel and entry and exit signals</p><p>It’s all well and good to show relative price channels and how they fit nicely</p><p>with general trend development; however, does it actually improve trading</p><p>efficiency? If the implementation of relative price channels is genuinely</p><p>effective, the testing process should show improved results as the parameters</p><p>are refined until, ultimately, the concept could be used as a stand-alone</p><p>technique. Whether any technique is applied as a stand-alone strategy is a</p><p>personal choice which will be gOverned to a degree by experience and</p><p>trading personality.</p><p>However, the ultimate test for any confirmation technique is whether it is</p><p>capable of producing a positive outcome on its own. Hence, the testing</p><p>parameters from the previous chapters will be refined and the concept tested</p><p>again, to see if trading efficiency is improved. The entry and exit signals</p><p>discussed above and 34 periods will be used to move beyond the benchmarks.</p><p>The benchmark performance for all confirmation indicators was comparable,</p><p>with the results being very similar. Therefore, it would be logical to expect a</p><p>similar change as each strategy is adapted to trend development, using 34</p><p>periods.</p><p>The first point of interest is to see whether the success rate can be</p><p>significantly lifted. The relative price channel performed poorly when applied</p><p>at 14 periods. If the indicator is now relative to trend development, there</p><p>should be a significant jump in the rate of profitable trades.</p><p>Beyond the benchmark — the relative price</p><p>channel at 34 periods</p><p>The test results for the relative price channel at 34 periods, also shown in</p><p>table 7.3, are as follows:</p><p>Table 7.3: relative price channel — 34 period</p><p>Profit summary</p><p>Profit status PROFITABLE</p><p>Starting capital $25 000.00</p><p>Finishing capital $86 169.61</p><p>Maximum equity/(date) $63 551.57(6/05/2005)</p><p>Minimum equity/(date) −$4762.69(28/09/2001)</p><p>Gross trade profit −$90 911.87(363.65%)</p><p>Gross trade loss −$29 742.25 (−118.97%)</p><p>Total net profit $61 169.61 (244.68%)</p><p>Average profit per trade $1422.55</p><p>Profit factor 3.0567</p><p>Profit index 67.28%</p><p>Total transaction cost $1720.00</p><p>Total slippage $0.00</p><p>Daily compound interest rate 0.0591%</p><p>Annualised compound interest rate 24.0852%</p><p>Trades processed 478</p><p>Trades taken 43</p><p>Partial trades taken 0</p><p>Trades rejected 435</p><p>Winning trades 19 (44.19%)</p><p>Losing trades 24 (55.81%)</p><p>Largest winning trade/(date) $20 293.49 (6/05/2005)</p><p>Largest losing trade/(date) −$2445.40 (26/05/2000)</p><p>Average winning trade $4784.84</p><p>Average losing trade −$1239.26</p><p>Average win/average loss 3.8610</p><p>Trade duration statistics (all trades)</p><p>Maximum trade duration 1596 (days)</p><p>Minimum trade duration 35 (days)</p><p>Average trade duration 281 (days)</p><p>Winning trades</p><p>Maximum trade duration 1596 (days)</p><p>Minimum trade duration 35 (days)</p><p>Average trade duration 393 (days)</p><p>Losing trades</p><p>Maximum trade duration 553 (days)</p><p>Minimum trade duration 35 (days)</p><p>Average trade duration 193 (days)</p><p>Consecutive trade statistics</p><p>Maximum consecutive winning trades 5</p><p>Maximum consecutive losing trades 5</p><p>Average consecutive winning trades 1.58</p><p>Average consecutive losing trades 2.18</p><p>Relative drawdown</p><p>Maximum dollar drawdown/(date) $6941.30 (28/09/2001)</p><p>Maximum percentage drawdown/(date) 25.6000% (28/09/2001)</p><p>Absolute (peak to valley) dollar drawdown</p><p>Maximum dollar drawdown $6941.30 (25.6000%)</p><p>Capital peak/(date) $27 118.61 (17/03/2000)</p><p>Capital valley/(date) $20 177.31 (28/09/2001)</p><p>The percentage of profitable trades has almost doubled to come in at an</p><p>acceptable 44.19 per cent, which is up considerably on the original 23.98</p><p>per cent produced by the benchmark result.</p><p>The annualised rate of return 24.08 per cent is an acceptable result, and</p><p>is up marginally on the benchmark return of 21.89 per cent. Not only has</p><p>the rate of profitable trades effectively doubled, the net return has also</p><p>improved slightly. It is clearly apparent that the revised channel is relative</p><p>to general trend development.</p><p>The peak to valley drawdown remains unacceptable at 25.60 per cent,</p><p>and this has deteriorated slightly on the benchmark result of 22.27 per</p><p>cent.</p><p>One of the biggest disappointments with the relative price channel is</p><p>the change with regard to the average loss. The benchmark produced an</p><p>average loss of just $233.47, which is very respectable; however, the 34-</p><p>period price channel generated an average loss of $1239.26. This tends to</p><p>suggest that there is an underlying issue with application in relation to</p><p>ongoing trend development.</p><p>The fact that the success rate of the original channel was doubled simply</p><p>changing the parameters from 14 to 34 and tinkering with the entry and exit</p><p>signals tends to suggest that the original values are largely irrelevant to</p><p>general trend development. This explains the high number of false signals</p><p>experienced by many. This is not to say that experienced traders cannot apply</p><p>indicators using a reduced number of periods. It simply suggests that the</p><p>degree of consistency is dramatically reduced, meaning an increased level of</p><p>experience is required in order to offset this negative impact.</p><p>Equity curve and yearly profits</p><p>Figure 7.4 shows the equity curve for the relative price channel at 34 periods.</p><p>Figure 7.4: closed equity curve for 34-period relative price channel</p><p>The equity curve has produced two periods of inactivity; however, the period</p><p>of primary concern commenced in late August 2004. The second pause</p><p>developed near the conclusion of the testing period, so too much emphasis</p><p>should not be placed on its appearance.</p><p>The solid rise throughout 2003 and into 2004 tends to suggest that a good</p><p>level of consistency was achieved in relation to legitimate confirmation</p><p>signals.</p><p>Figure 7.5 shows the yearly profits for the relative price channel at 34</p><p>periods.</p><p>Figure 7.5: yearly profits for 34-period relative price channel</p><p>The yearly return graph highlights consecutive losses for 2000 and 2001 of</p><p>12.68 per cent and 7.29 per cent respectively. Consistent returns were</p><p>produced, although it appears that the gains for 2002 and 2005 occurred Over</p><p>a relatively short period.</p><p>The legion of fans who apply the relative strength index at 14 periods may</p><p>be feeling as though I have taken this most sacred of indicators and shredded</p><p>it completely, which was not my intention. The aim of the discussion here is</p><p>to highlight the impact of irrelevant application, not to question the concept.</p><p>Also remember that the ongoing discussions are directed to position trading.</p><p>Whether you like to face facts or not, the following points have been</p><p>identified:</p><p>It can never be assumed that default values are necessarily relative to</p><p>trend development.</p><p>It can never be assumed that smoothing automatically enhances results.</p><p>Beyond the benchmark — internal strength</p><p>channel at 34 periods</p><p>As discussed previously, the conventional internal strength index is basically</p><p>a raw version of the relative strength index. The test results for the internal</p><p>strength channel at 34 periods, also shown in table 7.4, are as follows:</p><p>Table 7.4: internal strength channel — 34 period</p><p>Profit summary</p><p>Profit status PROFITABLE</p><p>Starting capital $25 000.00</p><p>Finishing capital $89 342.80</p><p>Maximum equity/(date) $69 431.66 (4/03/2005)</p><p>Minimum equity/(date) −$3830.78 (21/09/2001)</p><p>Gross trade profit $89 517.58 (358.07%)</p><p>Gross trade loss −$25 174.78 (−100.70%)</p><p>Total net profit $64 342.80 (257.37%)</p><p>Average profit per trade $748.17</p><p>Profit factor 3.5558</p><p>Profit index 71.88%</p><p>Total transaction cost $3440.00</p><p>Total slippage $0.00</p><p>Daily compound interest rate 0.0607%</p><p>Annualised compound interest rate 24.7778%</p><p>Trades processed 919</p><p>Trades taken 86</p><p>Partial trades taken 0</p><p>Trades rejected 833</p><p>Winning trades 46 (53.49%)</p><p>Losing trades 40 (46.51%)</p><p>Largest winning trade/(date) $13 616.56 (21/12/2001)</p><p>Largest losing trade/(date) −$2063.68 (20/05/2005)</p><p>Average winning trade $1946.03</p><p>Average losing trade −$629.37</p><p>Average win/average loss 3.0920</p><p>Trade duration statistics (all trades)</p><p>Maximum trade duration 661 (days)</p><p>Minimum trade duration 21 (days)</p><p>Average trade duration 184 (days)</p><p>Winning trades</p><p>Maximum trade duration 661 (days)</p><p>Minimum trade duration 28 (days)</p><p>Average trade duration 223 (days)</p><p>Losing trades</p><p>Maximum trade duration 490 (days)</p><p>Minimum trade duration 21 (days)</p><p>Average trade duration 140 (days)</p><p>Consecutive trade statistics</p><p>Maximum consecutive winning trades 11</p><p>Maximum consecutive losing trades 7</p><p>Average consecutive winning trades 2.56</p><p>Average consecutive losing trades 2.22</p><p>Relative drawdown</p><p>Maximum dollar drawdown/(date) $4154.92 (20/12/2002)</p><p>Maximum percentage drawdown/(date) 10.7900% (20/12/2002)</p><p>Absolute (peak to valley) dollar drawdown</p><p>Maximum dollar drawdown</p><p>$6028.85 (6.4010%)</p><p>Capital peak/(date) $94 191.66 (4/03/2005)</p><p>Capital valley/(date) $88 162.82 (12/08/2005)</p><p>The number of potential trading opportunities identified was 919,</p><p>which is a vast improvement Over the 3998 trades that were previously</p><p>identified using 14 periods. While this is still a sizeable number of</p><p>confirmation signals to be processed, the number is now manageable.</p><p>The percentage of successful trades comfortably broke the magical 50</p><p>per cent level, with the percentage of successful trades coming in at 53.49</p><p>per cent.</p><p>The peak to valley drawdown was a dramatic improvement — to the</p><p>point where I re-tested the concept and double-checked the coding for</p><p>errors. The worst-case scenario was a very acceptable 6.40 per cent, down</p><p>dramatically on the 23.34 per cent produced by the benchmark.</p><p>The average profit was $1946.03, which is up on the benchmark return</p><p>of $945.78.</p><p>The average loss has also increased from $240.73 to $629.37, which is</p><p>to be expected as I have continued with an exit with a close below the</p><p>upper cord of the bearish band.</p><p>The total net profit is up considerably on the benchmark return with a</p><p>net result of $64 342.80 or 257.37 per cent. The benchmark result was not</p><p>so appealing, at a very conservative $33 087.69 or 132.35 per cent.</p><p>Equity curve and yearly profits</p><p>Figure 7.6 shows the equity curve for the internal strength channel at 34</p><p>periods. The closed equity curve generally rose, peaking in March 2005. The</p><p>performance since then dipped slightly — although the testing period does</p><p>only include partial data for this year. While the strategy produced a flat</p><p>return during 2002, the positive aspect of this is that a significant loss was not</p><p>incurred during a bearish market. A positive aspect of the equity curve as a</p><p>whole is the minimal decline in value. There was only one notable dip during</p><p>late 2002; however, beyond this the drawdown on capital has been limited.</p><p>This tends to indicate that losses are cut relatively quickly and with a degree</p><p>of consistency. As expected, profits are allowed to develop. Again, while this</p><p>is not an ideal equity curve, it is one that you could live with as the</p><p>confirmation process is refined Over time.</p><p>Figure 7.6: closed equity curve for 34-period internal strength channel</p><p>Figure 7.7 shows the yearly profits for the internal strength channel at 34</p><p>periods.</p><p>Figure 7.7: yearly profits for 34-period internal strength channel</p><p>The yearly profits highlight solid growth in bullish conditions, especially for</p><p>2003, indicating that even when with what many would consider extremely</p><p>slow values for channel application, the ability to collect profits is not</p><p>diminished. The losses of 2000 and 2005 are disappointing, especially for</p><p>2005 as this year saw bullish activity; however, when these profits are</p><p>compared to the benchmark, it is clear that the results have been enhanced</p><p>significantly — simply by making the values relative to general trend</p><p>development.</p><p>Beyond the benchmark — directional</p><p>movement channel at 34 periods</p><p>In order to build on previous discussions, in this section I will include the</p><p>directional index at both 14 and 34 periods along with the directional</p><p>movement, or momentum bias, channel on price at 34 periods. So far I have</p><p>discussed the indicator values of 14 and 34 periods separately. Now that you</p><p>are starting to develop an understanding of channel application, I can</p><p>compare both values simultaneously in order to evaluate whether any</p><p>significant disparity in application exists. Figure 7.8 shows the directional</p><p>index at 14 and 34 periods. As you will notice in figure 7.8, the changes are</p><p>minimal at best when it comes to conventional indicator application.</p><p>The major peaks and troughs in indicator development remain unchanged</p><p>— contrary to what some would like us to believe, dispelling the myth that a</p><p>more responsive indicator is essential. I’m not going to suggest for one</p><p>moment that a short-term indicator when applied to swing trading is not a</p><p>superior concept; however, by and large, the majority of people who are</p><p>active within the market approach it from a position trading mindset, so for</p><p>these people, a super-responsive indicator is primarily a self-defeating</p><p>confirmation tool, as I will discuss later.</p><p>The 34-period directional index is slightly smoother than its short-term</p><p>cousin, while the divergence that later occurred following the parabolic</p><p>activity in 2003 was very similar in development. The key here is that, while</p><p>all major points of significance relative to price action and indicator</p><p>development remain unchanged, in the process many of the minor and</p><p>insignificant bumps that occurred along the way have been removed. Many</p><p>inexperienced traders delude themselves into believing that super-responsive</p><p>indicators are essential for early entry and ultimate success, automatically</p><p>assuming that the outcome will be positive. However, this is not necessarily</p><p>the case. As the test results have shown, most confirmation tools enjoy a</p><p>success rate of around 30 per cent when applied at 14 periods, meaning these</p><p>traders are living in a fool’s paradise.</p><p>Comparing 14 periods to 34 periods</p><p>Figure 7.8 shows the 14-period and 34-period directional movement index</p><p>and 34-period directional movement channel.</p><p>Figure 7.8: 14- and 34-period directional index and 34-period directional</p><p>movement channel</p><p>On figure 7.8, I have circled two areas of concern with the 14-period index.</p><p>The first region on the left suggests that there are early signs of a bearish</p><p>presence. However, looking at price action and the 34-period directional</p><p>index in both indicator and price channel format, can you see potential signs</p><p>of early bearish activity? Price action is consolidating as volatility decreases.</p><p>A trendline could also be placed below the consolidation period, although I</p><p>have left this out for clarity. There is little to suggest that a downtrend is</p><p>about to commence. What the 34-period index and channel show is sedate</p><p>price action that continues to consolidate.</p><p>The second area of concern, the circled area to the right, indicates</p><p>increasing strength as prices rise; however, this is not supported to the same</p><p>degree by its slower and more relative cousin. While the 14-period indicator</p><p>suggests that a rally is developing on increasing strength, the price channel</p><p>tends to suggest that the rally does not enjoy any notable bullish support. This</p><p>time around, a line of resistance could be placed above the highs of price</p><p>action. Logically, the best move would seem to be to wait for price action to</p><p>move into the bullish band along with a solid break above resistance before</p><p>committing.</p><p>While you can never eliminate false signals completely, the above shows</p><p>two points in indicator development that could have been applied as</p><p>confirmation for an entry, in either a long or a short position, that were later</p><p>proven to be inappropriate. These two confirmation signals could have been</p><p>easily avoided simply by applying values relative to trend development.</p><p>Before you pass this up by deciding that these values are too slow and you</p><p>will miss the bus, try to remember that the bus on route 14 later crashes three</p><p>times in four, so missing it is a blessing more often than it is a drawback. Try</p><p>to remember that short-term indicators are famous for false signals and little</p><p>else.</p><p>Another point of interest is the ‘divergence’ on the left of the 14-period</p><p>index. There is a decline in indicator values on static price action. Many</p><p>traditionalists would view this as a divergence and, while this may be</p><p>technically correct, how do you distinguish between the divergence on the</p><p>left, which was later followed by strong up-trending activity, and the</p><p>divergence on the right, which was later followed by a strong decline? While</p><p>it is not shown on the chart, the rallies that preceded each divergence both</p><p>increased by approximately 100 per cent and lasted for around 16 weeks.</p><p>With a very similar lead-up to each divergence, how can you separate the two</p><p>situations? Put simply, you cannot</p><p>do so with any confidence. Although</p><p>seasoned traders may be able to detect subtle changes that indicate probable</p><p>trend conclusion, the majority of less experienced traders see a divergence on</p><p>static price action.</p><p>Figure 7.9 shows a section of the chart shown in figure 7.8. This figure</p><p>shows the 34-period directional index and channel.</p><p>Figure 7.9: directional movement on price action</p><p>As figure 7.9 shows, directional movement has become relative to trend</p><p>development. This adaptation allows the following areas to be highlighted:</p><p>Throughout 2001, price action consistently tagged or moved beyond the</p><p>upper limits of the directional channel, indicating a strong bullish</p><p>presence underlying consolidating price action.</p><p>During 2002, price action consistently moved beyond the upper cord of</p><p>the bullish channel. The same process repeated itself as previously</p><p>mentioned, whereby price action tagged the lower cord of the bullish</p><p>channel before consolidating. The failure of price action to enter into the</p><p>neutral zone tends to indicate a strong bullish presence still supporting</p><p>price action.</p><p>From mid-2004, price action dropped into the neutral region, indicating</p><p>that the rally occurring at the time developed on neutral sentiment. The</p><p>rally during late 2004 did not enjoy strong bullish support. What is your</p><p>opinion of a rally that develops in the neutral zone?</p><p>In early 2005, price action declined into the bearish region with a clear</p><p>and strong downward move. Would there be any reason to hold this</p><p>stock from the long side? I would think not. Interestingly enough, the</p><p>bearish rally was very robust; however, it seldom moved beyond the</p><p>lower cord of the bearish region, indicating that the downward move has</p><p>an air of sustainability about it. Bearish price action seldom becomes</p><p>bullish, or vice versa, without first pausing in the process, usually in the</p><p>neutral zone.</p><p>How important do you think this type of information would be, if you were</p><p>considering a purchase? If you can identify a point of potential</p><p>unsustainability or a bullish rally in a predominantly bearish environment,</p><p>then, potentially, you have the ability to shift the odds of successful trading in</p><p>your favour. Being able to actually plot the indicator on price action allows</p><p>you to observe the impact that a range of values may have and the influence</p><p>of smoothing on the Overall outcome.</p><p>Test results</p><p>The test results for the directional movement channel at 34 periods, also</p><p>shown in table 7.5, are as follows:</p><p>Table 7.5: directional movement channel — 34 period</p><p>Profit summary</p><p>Profit status PROFITABLE</p><p>Starting capital $25 000.00</p><p>Finishing capital $96 086.33</p><p>Maximum equity/(date) $71 086.33(7/10/2005)</p><p>Minimum equity/(date) $3024.44(11/02/2000)</p><p>Gross trade profit $90 290.08(361.16%)</p><p>Gross trade loss −$19203.75 (−76.82%)</p><p>Total net profit $71 086.33 (284.35%)</p><p>Average profit per trade $1077.07</p><p>Profit factor 4.7017</p><p>Profit index 78.73%</p><p>Total transaction cost $2640.00</p><p>Total slippage $0.00</p><p>Daily compound interest rate 0.0641%</p><p>Annualised compound interest rate 26.3660%</p><p>Trades processed 631</p><p>Trades taken 66</p><p>Partial trades taken 0</p><p>Trades rejected 565</p><p>Winning trades 42 (63.64%)</p><p>Losing trades 24 (36.36%)</p><p>Largest winning trade/(date) $12 079.38(28/11/2003)</p><p>Largest losing trade/(date) −$1933.75(10/09/2004)</p><p>Average winning trade $2149.76</p><p>Average losing trade −$800.16</p><p>Average win/average loss 2.6867</p><p>Trade duration statistics (all trades)</p><p>Maximum trade duration 770 (days)</p><p>Minimum trade duration 28 (days)</p><p>Average trade duration 269 (days)</p><p>Winning trades</p><p>Maximum trade duration 770 (days)</p><p>Minimum trade duration 28 (days)</p><p>Average trade duration 329 (days)</p><p>Losing trades</p><p>Maximum trade duration 735 (days)</p><p>Minimum trade duration 28 (days)</p><p>Average trade duration 163 (days)</p><p>Consecutive trade statistics</p><p>Maximum consecutive winning trades 11</p><p>Maximum consecutive losing trades 6</p><p>Average consecutive winning trades 3.00</p><p>Average consecutive losing trades 1.85</p><p>Relative drawdown</p><p>Maximum dollar drawdown/(date) $3861.58 (13/05/2005)</p><p>Maximum percentage drawdown/(date) 6.9640% (30/05/2003)</p><p>Absolute (peak to valley) dollar drawdown</p><p>Maximum dollar drawdown $5200.52 (5.5280%)</p><p>Capital peak/(date) $94 071.24 (18/03/2005)</p><p>Capital valley/(date) $88 870.72 (3/06/2005)</p><p>The success rate for profitable trades was a most impressive 63.64 per</p><p>cent, which is Over double the 30.14 per cent produced by the benchmark.</p><p>The peak to valley drawdown, or the worst-case scenario, has also</p><p>improved, going down from 26.83 per cent to a very respectable 5.52 per</p><p>cent.</p><p>The average loss has blown out to $800.16, which stemmed from a</p><p>lower exit point. With the benchmark, both entry and exit points were</p><p>identified using the lower cord of the bullish band as this more closely</p><p>resembled general market application and the belief that an index should</p><p>not go below 50 per cent. By lowering the exit point to the upper cord of</p><p>the bearish band, trades were not only opened up to the possibility of</p><p>absorbing consolidation activity, they were also exposed to additional</p><p>downside when things didn’t work out as expected.</p><p>On the other side of the coin, the average profit has improved slightly</p><p>from $2055.02 to $2149.76, which has possibly stemmed from the</p><p>collection of profits following extreme trending activity.</p><p>The benchmark produced a total net profit of $87 246.14 (or 348.98 per</p><p>cent), which is not too shabby considering the opening account balance of</p><p>just $25 000.00. The relative approach produced a notably lower result by</p><p>returning a total net profit of $71 086.33 (or 284.35 per cent) Over the</p><p>testing period. To put this another way, the benchmark produced an</p><p>annualised net return of 30.16 per cent while the relative version produced</p><p>the slightly more conservative figure of 26.36 per cent.</p><p>The biggest single change related to the number of trades identified.</p><p>While the benchmark or conventional approach that used 14 periods</p><p>identified 2863 potential opportunities, the relative version identified 631,</p><p>which is far more acceptable for real-time application.</p><p>Equity curve and yearly profits</p><p>Figure 7.10 shows the equity curve for the 34-period momentum bias</p><p>channel.</p><p>Figure 7.10: equity curve for 34-period directional movement channel</p><p>As figure 7.10 shows, while the closed equity curve for the relative approach</p><p>is far from perfect, it is a vast improvement Over the original, showing</p><p>consistent growth since 2000 with the exception of the last half of 2002 and</p><p>mid-2005.</p><p>Figure 7.11 shows the yearly profits for the 34-period directional</p><p>movement channel.</p><p>Figure 7.11: yearly profits for 34-period directional movement channel</p><p>The yearly profit graph indicates a relatively consistent return, although 2004</p><p>was fairly conservative considering strong gains were made by the broader</p><p>market. While 2003 produced an above-average result, this is to be expected</p><p>considering the robust bullish activity that developed following the March</p><p>low.</p><p>Conclusions so far</p><p>Regardless of whether you choose to trade relative price channels as a stand-</p><p>alone approach or in conjunction with specialised entry triggers, you should</p><p>know whether you have a confirmation tool that is effective in its own right. I</p><p>know that many of you will be shaking your heads in either disbelief or</p><p>disagreement; however, I suggest that you take the time to test your strategy</p><p>with and then without your confirmation tools. Don’t limit yourself to just</p><p>testing your entry trigger and the trailing stop; also test your confirmation</p><p>tool as you are relying on this for effective confirmation. Then combine both</p><p>as a complete strategy to assess any changes. While you may automatically</p><p>assume that the best results are achieved when you combine both, this is</p><p>seldom the case, especially on a daily time frame, and you may be very</p><p>disappointed with the results. If a confirmation tool cannot test profitable or</p><p>reflect a degree of consistency, it is not a confirmation technique — it is a</p><p>confidence tool.</p><p>The vast majority of inexperienced traders use confirmation tools yet have</p><p>no</p><p>idea of whether they are truly effective. I often hear comments like, ‘I use</p><p>the MACD-H or the Stochastic and it works a real treat’. How do they know</p><p>that it works a real treat if they do not know how their strategy performs</p><p>without it, or how the indicator tests on its own or at a range of alternative</p><p>parameters? Perhaps they assume that it is an effective confirmation tool</p><p>because, as a profitable outcome was achieved, the indicator somehow</p><p>contributed to the end result. They may find that their favourite ‘whizzbang’</p><p>confirmation indicator is more like ‘fizz pop’ if they took the time to analyse</p><p>a little more deeply.</p><p>Results at this early stage tend to suggest that the relative approach is</p><p>exactly that — relative. This is born out by the Overall improvement made by</p><p>confirmation efficiency. For the strategy to produce a successful outcome on</p><p>63.64 per cent of all trades without additional confirmation puts beyond any</p><p>doubt whether relative analysis is effective when it comes to the adaptation of</p><p>directional movement. Accepted practice based on lunar cycles is probably</p><p>not that appropriate for the position trader in today’s environment. Before</p><p>you start thinking, ‘I like the thought of a 30 per cent return’, also remember</p><p>that the odds of success are about the same. You have a choice — a 30 per</p><p>cent return driven by a 30 per cent probability of success, or a 26 per cent</p><p>average annual return with the probability of a 63 per cent success rate. You</p><p>do the sums.</p><p>Beyond the benchmark — volume bias</p><p>channel at 34 periods</p><p>One of the most neglected of all tools is volume. While single day behaviour</p><p>may tell you little, collective volume analysis can be a very revealing tool,</p><p>especially for more savvy traders.</p><p>Figure 7.12 shows the volume bias index and channel at 34 weeks.</p><p>Figure 7.12: relative volume bias index and channel</p><p>Looking at figure 7.12, do you think you could differentiate between a bullish</p><p>and bearish volume bias now?</p><p>Remembering that the picture received when applying volume to price</p><p>action will differ when compared with momentum-based indicators, the</p><p>following points can be highlighted from figure 7.12:</p><p>Throughout 2002 EQI developed on a strong bullish bias in volume</p><p>activity. With price action consistently in the upper channel, the volume</p><p>associated with higher closing prices is consistently larger than the</p><p>volume associated with any sell-off. It is unlikely that any bearish</p><p>activity will develop while volume continues to favour higher closing</p><p>prices.</p><p>The parabolic rally in 2003 developed on strong bullish volume, as you</p><p>would expect. Naturally, any bearish volume associated with the</p><p>parabolic rally was minimal at best — otherwise, parabolic activity is</p><p>unable to occur.</p><p>Following the rally, price action finally punched beyond the upper cord of the</p><p>bullish band before stalling. While the bullish presence is very strong, as</p><p>indicated by the relationship that exists between price action and the volume</p><p>channel, price action has not moved. From this you could conclude that,</p><p>while the buyers have met an equal number of sellers and are no longer able</p><p>to bid prices higher, the number of sellers present in the market has only</p><p>managed to stall the trend, not reverse it at this stage. (I have repeated this</p><p>observation to reinforce the notion of volume analysis and trend</p><p>development, not momentum. It can take some time to differentiate between</p><p>the two if expectations are based on the previous application of momentum-</p><p>based indicators.)</p><p>Limits on volume activity can now be defined. Just how long can a bullish</p><p>presence remain dominant and at what point does the stranglehold on price</p><p>action become unsustainable? Until now, identifying unsustainable price</p><p>action has not been possible, simply because volume has never been adapted</p><p>to price action in this particular manner — that I am aware of anyway. The</p><p>moment price action penetrates the outer extremities of the volume channel,</p><p>the trend either pauses or retraces. This suggests that, like all other actions</p><p>within the market, there comes a point of unsustainability. The relationship</p><p>between price, volume and trend development can now be quantified. Finally</p><p>you can place expectancy on volume-driven trend development.</p><p>This can be taken one step further. Once price action moves beyond the</p><p>outer extremities of the volume channel, it has potentially reached a point that</p><p>can no longer be supported by the current levels of supply and demand. An</p><p>extreme bullish value occurs because of significant volume activity that</p><p>distinctly favours one course of action. In order to sustain an extreme bullish</p><p>value, the current buying activity must continue while the sellers stay away.</p><p>As discussed, as trends develop volume decreases for two reasons — the</p><p>large traders have already committed, while the ‘mum and dad’ investors lack</p><p>the necessary capital to drive the market. The essential ingredient required for</p><p>trend continuation is a depleting resource and that depleting resource is the</p><p>same fuel that started the trend in the first place — buyers.</p><p>Figure 7.13 shows the volume bias index and channel at 34 periods,</p><p>adapted to the price action of Gunns (GNS).</p><p>Figure 7.13: volume bias and bearish activity</p><p>When applying a relative price channel, regardless of whether it incorporates</p><p>volume, momentum or volatility, I analyse the relationship between price</p><p>action, the conventional indicator and the channel. I continue to apply the</p><p>conventional indicator format in conjunction with the price channel so I can</p><p>evaluate where price action sits in relation to channel development from an</p><p>alternative perspective.</p><p>I use the bottom indicator to define the relationship that exists between</p><p>price and volume, and I use the channel to define the relationship that exists</p><p>between indicator and price. In the end, I have a complete picture relative to</p><p>trend development. The channel is relative to price action, ensuring that</p><p>volume is relative to trend development. As volume is relative to both the</p><p>channel parameters and price action, the circle of relative analysis is complete</p><p>— and I have used price action to determine what is relative, not the phase of</p><p>the moon.</p><p>Looking at figure 7.13, you can quickly see that the trend of 2003</p><p>developed on a bullish presence, as it should in order for the trend to develop</p><p>in the first place. However, the volume behind higher closes was not</p><p>significant when compared to the volume driving lower closing prices. With</p><p>the volume bias index consistently hOvering around the 55 per cent to 60 per</p><p>cent region, it is clear that the bullish volume was only dominating trading</p><p>activity by 10 per cent to 20 per cent.</p><p>In 2005, price action declined as the bullish presence largely disappeared.</p><p>Volume activity plateaued to the point of equilibrium where the level of</p><p>liquidity driving higher closing prices was comparable with the volume</p><p>associated with lower closing prices.</p><p>From May 2005, the volume associated with lower closing prices began to</p><p>dominate and price action passed through the lower band, indicating a solid</p><p>swing toward bearish volume activity.</p><p>If volume is truly irrelevant to general trend development (as some</p><p>promote), there should not be any notable improvement Over the benchmark</p><p>when the volume bias channel is evaluated at 34 periods. Try to remember</p><p>that to be even considering this type of an approach to volume analysis is</p><p>usually considered impossible by the majority.</p><p>Test results</p><p>The test results for the volume bias channel at 34 periods, also shown in table</p><p>7.6, are as follows:</p><p>Table 7.6: volume bias channel — 34 period</p><p>Profit summary</p><p>Profit status PROFITABLE</p><p>Starting capital $25 000.00</p><p>Finishing Capital $108 836.74</p><p>Maximum equity/(date) $85 433.91 (7/10/2005)</p><p>Minimum Equity/(Date) −$3 703.72(28/09/2001)</p><p>Gross Trade Profit $103 944.69(415.78%)</p><p>Gross Trade Loss −$20 107.95 (−80.43%)</p><p>Total Net Profit $83 836.74 (335.35%)</p><p>Average Profit Per Trade $1552.53</p><p>Profit Factor 5.1693</p><p>Profit Index 80.66%</p><p>Total Transaction Cost $2160.00</p><p>Total</p><p>Slippage $0.00</p><p>Daily Compound Interest Rate 0.0696%</p><p>Annualised Compound Interest Rate 28.8986%</p><p>Trades Processed 642</p><p>Trades Taken 54</p><p>Partial Trades Taken 0</p><p>Trades Rejected 588</p><p>Winning Trades 30 (55.56%)</p><p>Losing Trades 24 (44.44%)</p><p>Largest Winning Trade/(Date) $26 690.66 (28/02/2003)</p><p>Largest Losing Trade/(Date) −$2769.87(16/10/2002)</p><p>Average Winning Trade $3464.82</p><p>Average Losing Trade −$837.83</p><p>Average Win/Average Loss 4.1355</p><p>Trade Duration Statistics (All Trades)</p><p>Maximum Trade Duration 994 (Days)</p><p>Minimum Trade Duration 34 (Days)</p><p>Average Trade Duration 322 (Days)</p><p>Winning Trades</p><p>Maximum Trade Duration 994 (Days)</p><p>Minimum Trade Duration 34 (Days)</p><p>Average Trade Duration 418 (Days)</p><p>Losing Trades</p><p>Maximum Trade Duration 511 (Days)</p><p>Minimum Trade Duration 55 (Days)</p><p>Average Trade Duration 202 (Days)</p><p>Consecutive Trade Statistics</p><p>Maximum Consecutive Winning Trades 8</p><p>Maximum Consecutive Losing Trades 5</p><p>Average Consecutive Winning Trades 2.73</p><p>Average Consecutive Losing Trades 2.00</p><p>Relative Drawdown</p><p>Maximum Dollar Drawdown/(Date) $4248.95 (26/11/2004)</p><p>Maximum Percentage Drawdown/(Date) 13.0500% (28/09/2001)</p><p>Absolute (Peak To Valley) Dollar Drawdown</p><p>Maximum Dollar Drawdown $4248.95 (4.8410%)</p><p>Capital Peak/(Date) $87 762.86 (18/06/2004)</p><p>Capital Valley/(Date) $83 513.91 (26/11/2004)</p><p>The number of potential candidates was reduced to the more</p><p>manageable size of 642, which is down significantly from the 4024</p><p>created by the benchmark.</p><p>The average profit did not change significantly, although it was still an</p><p>improvement on the benchmark, concluding at $83 836.74, which equates</p><p>to an annualised return of 28.89 per cent.</p><p>The maximum peak to valley drawdown has also improved by</p><p>declining to an average of 4.84 per cent, down from the 18.13 per cent</p><p>produced by the 14-period channel.</p><p>One of the biggest improvements came in consistency. The success rate</p><p>of profitable trades lifted from 35.29 per cent to 55.56 per cent, which has</p><p>no doubt contributed to the general improvement experienced by other</p><p>elements of the technique.</p><p>Again, there is little doubt that when techniques that are relative to trend</p><p>development are applied, regardless of whether they are price or volume</p><p>based, an improvement in consistency is achieved. There have been positive</p><p>changes to all facets of the volume channel, because the parameters have</p><p>been adapted to coincide with general trend development. Keeping in mind</p><p>that the entry and exit signals are solely reliant on volume behaviour, it</p><p>brings into question comments made by some that volume doesn’t count,</p><p>especially in the Australian market due to the low level of liquidity. While it</p><p>may not be the key to trading riches, another major piece of the trading</p><p>puzzle has just slotted into place.</p><p>Equity curve and yearly profits</p><p>Figure 7.14 shows the equity curve for the 34-period volume bias channel.</p><p>Figure 7.14: equity curve for 34-period volume bias channel</p><p>Looking at figure 7.14, can you believe that you are even considering an</p><p>equity curve for volume? The equity curve has developed in an up-trending</p><p>manner, albeit slightly inconsistently. The surge from late 2002 preceded the</p><p>March 2003 reversal, meaning there was a rise in bullish volume activity</p><p>prior to the trend change defined by the major indices. The surge consisted of</p><p>numerous trades Over many weeks.</p><p>Figure 7.15 shows the yearly profits from the volume bias channel strategy</p><p>at 34 periods. The following points can be made:</p><p>Figure 7.15: yearly profits for 34-period volume bias channel</p><p>Bullish volume activity plateaued throughout 2004, which led to a</p><p>reduced number of signals being identified. Given the maturity of the</p><p>market and that the majority of large investors would have been fully</p><p>exposed during this period, a return of 20.34 per cent is acceptable.</p><p>In 2005, a lift in trading activity coincided with an increased result.</p><p>Although the bulk of the gains stemmed from the conclusion of the</p><p>March retracement, which is confirmed by the equity curve, a half-year</p><p>result of 27.85 per cent would have to considered more than acceptable.</p><p>A loss of 12.35 per cent for 2001 is definitely undesirable; however,</p><p>remember that in real time this style of volume analysis would be used</p><p>in conjunction with a trend-based entry trigger.</p><p>What can be determined from the equity curve and yearly profits is the</p><p>technique’s ability to capture trend reversals. Although the market reversed in</p><p>March 2003, bullish volume began to filter into price action in late 2002. At</p><p>the time, the majority were unaware that this was occurring. However, the</p><p>movement is highlighted by the significant jump in the equity curve and</p><p>yearly return in 2002.</p><p>The first notable retracement then occurred in March 2005; however, as</p><p>volume was rising prior to this event, it could be assumed that capital was</p><p>flowing from one sector of the market to another. Once the volume activity</p><p>paused, the market dropped. Can you guess which stocks retraced? The</p><p>retracement came about because of the satisfactory conclusion of transferred</p><p>capital. With a percentage of the market no longer supporting those stocks</p><p>that have since been sold off, a retracement is inevitable. Interesting enough,</p><p>the transfer of capital could be seen well before the ensuing retracement</p><p>occurred.</p><p>Important exclusion</p><p>One trade was excluded from the 34-period volume bias channel testing</p><p>process. SDI produced a 2557.58 per cent return Over a four-and-a-half-year</p><p>period (6 January 2000 to 17 July 2004) with an entry price of $0.066 and an</p><p>exit at $1.75. I decided that this trade was potentially a disproportionate</p><p>representation of the effectiveness of relative volume analysis, and so I</p><p>deleted it from the results. While it is quite possible that a position trader may</p><p>have entered this particular trade at the time in question, it is questionable</p><p>whether the majority would have held the position for this period of time. As</p><p>SDI was not identified by other channel techniques, I considered it acceptable</p><p>to exclude it from testing results.</p><p>Relative analysis and compatibility</p><p>Does relative analysis work? I will let you be the judge. By applying</p><p>accepted confirmation techniques directly to price through the application of</p><p>a price channel, parameters that more clearly define and coincide with</p><p>general trend development can be applied. In this chapter, the retesting of</p><p>what is now relative application produced a solid improvement across all</p><p>facets of the analysis process for all techniques, as the success rate using the</p><p>relative techniques generally doubled Over the rate of the benchmarks and</p><p>profits improved.</p><p>The test results show that applying relative values is far more effective than</p><p>applying popular or accepted parameters. Some may persist with accepted</p><p>practice and default values; however, there is little doubt that what is popular</p><p>is not always relative.</p><p>You now have the capacity to evaluate the compatibility of indicator</p><p>parameters against general trend development. Is it possible to develop and</p><p>implement stand-alone strategies using only relative price channels?</p><p>The next two chapters will begin to explore the bands in even greater detail,</p><p>focusing on the following two areas:</p><p>Confirmation — if you have the ability to define compatibility more</p><p>clearly, you can apply a relative price channel that is relative to general</p><p>trend development as a means of confirmation for existing strategies.</p><p>Bilateral relationship — now that you have the ability to Overlay</p><p>indicators directly on price action, you can also look for behaviour that</p><p>can lead to outcomes that you are currently unaware of.</p><p>I have not repeated the formulas for relative price channels as discussed</p><p>throughout this chapter as the parameters were simply set at 34 periods as a</p><p>part of the coding process. This will automatically apply the channels to price</p><p>action, as discussed in this chapter.</p><p>Chapter 8</p><p>Confirmation</p><p>So far I have discussed the application of price channels as if traded as a</p><p>stand-alone technique. I am not arguing that relative price channels should</p><p>be</p><p>used as real-time stand-alone strategies — although they could be by more</p><p>experienced traders — but rather that the impact of applying relative analysis</p><p>or the impact of applying indicator values needs to be determined. The</p><p>relative analysis applied must be relevant to trend development. In reality,</p><p>very few people would use price channels as a stand-alone tool — we all</p><p>have our preferred techniques. I look for complex turning points and initial</p><p>highs.</p><p>Tip</p><p>Confirmation analysis is an integral part of the Overall process; however, in order for it to</p><p>be effective it must also be relative. Few traders would apply a short-term moving average</p><p>of just a few periods for trend identification, yet many will apply short-term confirmation</p><p>indicators in an attempt to identify trends from an alternative perspective.</p><p>Channels combined with initial highs</p><p>Once you have identified compatible parameters for the price channel of your</p><p>preference, you can use the channel as a legitimate confirmation tool for</p><p>existing techniques. Let’s say that you trade complex true turning points. You</p><p>can now test whether your existing strategy can be improved upon — for</p><p>example, by taking only long side positions when a bullish complex turning</p><p>develops within the bullish band. Alternatively, you could consider shorting</p><p>— for example, if the stock experiences a bearish true turning point within</p><p>the bearish band. By trading existing strategies relative to the band price</p><p>action is occupying, you minimise the chance of trading against the trend.</p><p>So far, your trading approach is as follows:</p><p>If price action is in the bullish band, trade from the long side.</p><p>If price action is in the bearish band, trade from the short side.</p><p>The approach is now starting to come together. The preferred indicators have</p><p>been adapted to price action in order to develop relative price channels. Now</p><p>that a relative relationship between indicator values and trend development</p><p>has been defined from a more general perspective, you can apply a specific</p><p>entry strategy using the price channels.</p><p>As many are aware, one of my preferred techniques involves trading initial</p><p>highs. If you are having any doubts as to whether relative analysis is pertinent</p><p>as a confirmation tool, the following discussion should put any doubts to rest.</p><p>In order to highlight the relevance of relative analysis I tested the initial high</p><p>technique without a means of confirmation in order to create a benchmark for</p><p>further comparison. I then applied the classic 14-period indicator as used by</p><p>the majority. Finally, I applied parameters relative to trend development.</p><p>Through the process you will be able to observe the changes that take place.</p><p>Before I start testing the effectiveness of the relative price channel, I will</p><p>first explain the initial high entry trigger. There is little point of explaining</p><p>only one half of a strategy — it is hard to see how effective an exit is if you</p><p>do not have an understanding of the entry set-up.</p><p>Initial highs</p><p>Trading involves developing an idea on how to profit from a change in price</p><p>action. Some ideas exceed all expectations, while some are quickly forgotten</p><p>— well, many are quickly forgotten actually. The initial high technique</p><p>evolved Over time due to my ongoing dissatisfaction with trading new highs.</p><p>The principle of trading a new high in any trend is sound, as it requires the</p><p>purchase of a stock in a rising trend or rally; however, it can lead to the trader</p><p>chasing price action unnecessarily if a definitive entry point is not determined</p><p>prior to trade assessment.</p><p>Trading 21- and 30-day highs is a very common approach, but is this</p><p>universally accepted technique compatible with general market behaviour?</p><p>The process usually involves traders identifying the highest high price within</p><p>the last 21 days and then designing an entry process around this occurrence.</p><p>The most common method of application is that if a new 21-day high is</p><p>experienced, this becomes the entry trigger. Traders will enter the following</p><p>day in accordance with their trading plans.</p><p>I have two concerns with this type of approach. First up, I find any trading</p><p>strategy that is based on the principles of new high development too loose in</p><p>its construction — it needs to also define what constitutes an entry point and</p><p>what does not. Added to this, who said that 21- or 30-day highs were the</p><p>optimum values for trading? They may work sometimes, but are these values</p><p>the optimum periods for application under a range of conditions? For</p><p>continuity, I will build this discussion around what is accepted practice and</p><p>use an initial 21-day high.</p><p>Initial 21-day highs</p><p>Naturally, the initial 21-day high is the first 21-day high in the trend. I then</p><p>ignore all following 21-day highs until a new initial 21-day high is generated.</p><p>Every new high that occurs following the initial 21-day high is in it own right</p><p>a new 21-day high; however, does this mean that every new high produced is</p><p>a legitimate entry trigger? If a rally runs Over a four- or five-week period,</p><p>there could be many 21-day highs. While some will argue that many</p><p>opportunities are missed by solely focusing on the initial high, I do not like</p><p>chasing price action. I focus on entry triggers, as much capital can be wasted</p><p>through chasing price action and not applying a clearly defined technique.</p><p>Figure 8.1 shows the price action for EQI during an uptrend. I have</p><p>highlighted on the chart the bulk of the new highs during the uptrend. At a</p><p>minimum they are all 21-day highs, but how many of the new highs actually</p><p>represent a legitimate entry point?</p><p>Figure 8.1: initial high concept</p><p>The repeated problem I encountered was the indecisive nature of the new</p><p>high signal. Inexperience led to me chasing price action at the appearance of</p><p>a new high. I had repeatedly read that trading new highs was a logical</p><p>approach and, even with my limited experience at the time, I could see the</p><p>importance of buying into rising price action — especially as bottom fishing</p><p>was not proving to be a very lucrative fishing ground. What I failed to realise</p><p>was that I needed to put the new high into the context of the surrounding</p><p>trend development. This is appropriate for any strategy regardless of the entry</p><p>trigger. The entry signal must be put into the context of the surrounding price</p><p>action in order to determine its significance. I needed a new high that said to</p><p>one L. Wilson, ‘This is an entry trigger, buy now’. If I was going to trade 21-</p><p>day highs, I should trade on the first 21-day high that occurs.</p><p>This line of thought eventually led to the initial high technique as used</p><p>today. To identify the first 21-day high, you first need to identify declining</p><p>price action and then plot 21 days forward of each price bar. Once price</p><p>actions stalls, the initial high reference point remains static until there is a</p><p>close above this point. If price action declines for less than 21 days, the</p><p>calculation process resets itself and recommences its decline from the most</p><p>recent high. Until a previous high is confirmed as the new reference for future</p><p>price action, the initial high indicator holds its previous value. The coding for</p><p>BullCharts and MetaStock is written so the indicator only plots once a</p><p>decline of 21 days or more is underway, as shown in figure 8.1.</p><p>According to accepted practice, simply trading new highs is sound. While</p><p>the principles of the concept are generally sound and while I mean no</p><p>disrespect to those traders who advocate the concept, I cannot agree with the</p><p>belief that the mere appearance of a new high constitutes an entry signal. The</p><p>strong point of trading new highs is the high probability of buying into rising</p><p>price action. Naturally, the concept of trading new highs is trend specific, so</p><p>the concept loses validity as the trend matures. Price action repeatedly makes</p><p>new highs as the trend matures before eventually retracing. Rallies and trends</p><p>are driven by the balance of supply and demand, whereby demand is at its</p><p>strongest when prices are lowest. Traders become increasingly</p><p>hesitant to</p><p>purchase as prices rise, while sellers are more willing to capitalise on higher</p><p>values. This causes the pendulum to swing in favour of supply.</p><p>It’s this constant ebb and flow between supply and demand that generates</p><p>trends and retracements. If the strength in trading new highs diminishes as the</p><p>trend matures due to the inability of price action to constantly produce new</p><p>highs brought about by a swing toward a domination of supply, then the</p><p>strongest new high signal would be among the first entry signals — those</p><p>generated at the start of the trend when demand is potentially at its strongest.</p><p>Plotting the initial 21-day high</p><p>In order to minimise any confusion with regard to the indicator re-plotting</p><p>itself following a recent high in price action, I will break down the technique</p><p>in a little more detail. Let’s continue with a 21-period high for continuity. I</p><p>have applied this to a weekly chart for greater clarity. In order for the initial</p><p>high indicator to re-plot above price action, price action must first experience</p><p>20 consecutive weeks of lower highs than the high of 21 weeks ago. This</p><p>must occur in order to identify the first new high.</p><p>The basic rules that determine how the indicator plots are as follows:</p><p>The indicator can only plot above declining price action.</p><p>The indicator will only re-plot once declining price action Over a 20-</p><p>week period is less than the high of 21 weeks ago.</p><p>Figure 8.2 shows the price action of AMP during a downtrend and how the</p><p>initial high indicator plotted during this time.</p><p>Figure 8.2: initial high indicator — plotting during down trends</p><p>Figure 8.3 shows an expanded view of the high (circled) in figure 8.2. In</p><p>figure 8.3, the indicator is referencing the high of 21 weeks ago; however, it</p><p>plots above the 20 weeks of declining price action that follows the high of 21</p><p>weeks ago. This means the indicator does not commence at the very high of</p><p>the trend but at one bar from the high.</p><p>Figure 8.3: initial high indicator and stalled price action</p><p>The indicator effectively references two periods in trend development</p><p>simultaneously, as follows:</p><p>The indicator references a prominent high, which in this case is 21</p><p>periods ago. It uses this position as the starting position for future</p><p>reference in regard to declining price action.</p><p>It also references the ensuing 20 weeks of price action following the</p><p>prominent high, ensuring that the 20 weeks of subsequent price action is</p><p>lower than the prominent high.</p><p>Only when the above criteria are met will the indicator plot above declining</p><p>price action. As shown in figure 8.3, the indicator commenced plotting</p><p>having confirmed 20 consecutive weeks of equal or lower price action. As the</p><p>downtrend continued the initial high referenced the high of 21 weeks prior,</p><p>while progressively moving forward. As long as price action continued to</p><p>produce 20 consecutive weeks of equal or lower trading activity, the process</p><p>of plotting a lower value continued. Price action will usually pause at some</p><p>point before either reversing — and also creating a potential entry signal —</p><p>or continuing with previous down-trending activity. The perception on how</p><p>the indicator plots is usually based on more conventional new high</p><p>techniques, which usually involve the indicator referencing the high. This is</p><p>not the case with the initial high indicator. It first identifies 21 periods of</p><p>lower trading activity in order to identify an early appearance of new high</p><p>activity.</p><p>In figure 8.3, the second high pokes its head above the initial high</p><p>indicator, so many people would expect to see a rise in indicator value. When</p><p>they don’t, they immediately assume there is a plotting error which is</p><p>incorrect. The 20 weeks of price action that has followed high 2 has been</p><p>lower, so the indicator recommences its decline using high 2 as the new</p><p>prominent point for reference. It unfolds in exactly the same manner as for</p><p>high 1. The indicator can never plot a higher value as it’s designed to assist in</p><p>identifying the first high in early trend development following a decline, so it</p><p>holds its previous value following high 2 or until new bearish conditions are</p><p>met.</p><p>For those interested in the high two periods earlier, high 3 is ignored as it</p><p>does not experience 20 weeks of declining price action immediately</p><p>thereafter.</p><p>Break-outs after the specified number of periods</p><p>It has been suggested that, while the indicator is designed to identify an initial</p><p>high Over a specified number of periods, this does not always occur. The</p><p>indicator is required to step away from price action as the downtrend unfolds</p><p>so that the first high that occurs above the initial indicator will be correct;</p><p>however, this does not imply that a 21-week indicator only identifies initial</p><p>21 week highs. It identifies initial 21-week highs at a minimum, but the actual</p><p>break-out may occur some weeks later. The price action for AMP, shown in</p><p>figure 8.4, highlights this issue.</p><p>Figure 8.4: initial high indicator during consolidation periods or trend</p><p>conclusion</p><p>As figure 8.4 shows, the initial high indicator plotted above declining price</p><p>action. The downtrend finally concluded and then drifted into what could</p><p>loosely be described as a consolidation period. The indicator commenced</p><p>plotting a static value but it took 24 weeks for price action to plot a new high.</p><p>This also became the prominent high and a new reference point for the</p><p>indicator in 21 weeks’ time should a break-out fail to develop. Price action</p><p>continued to drift for another 20 weeks before eventually closing higher,</p><p>effectively nullifying the previous high. With a new high having now</p><p>developed, followed by a steady rise in price action, the indicator patiently</p><p>waits for a new 20-week decline following a prominent high.</p><p>Some will see 44 weeks and argue this is not a 21-week high. In fairness, if</p><p>this was a more conventional technique for identifying new highs, such</p><p>comments would probably be justified; however, my aim is to identify the</p><p>early signs of trend development and in this aim, the technique performs as</p><p>expected. Just a pity it’s AMP.</p><p>Initial high indicator and repetitive entry signals</p><p>Figure 8.5 shows the price action of Newcrest Mining (NCM) and the initial</p><p>high indicator. This stock was brought to my attention by a couple of readers,</p><p>who questioned the validity of the technique due to repetitive entry signals.</p><p>The price action of NCM initially broke above the initial high indicator on 12</p><p>August 2005, only to immediately retreat. It then closed above the initial high</p><p>value for a second time on 2 September 2005, creating a second signal.</p><p>Readers asked how such an occurrence should be managed, when it was</p><p>perceived that this represented incorrect application.</p><p>Figure 8.5: initial high indicator with multiple break-outs</p><p>The first mistake is to assume that all break-outs will occur in a clinical,</p><p>decisive manner and that every signal will be singular in occurrence — and</p><p>that applies to any strategy. This kind of behaviour will occur on a semi-</p><p>regular basis, but this does not detract from the appearance or strength of</p><p>developing price action. As I have said previously, try to understand the</p><p>concept rather than just the trigger. The concept involves identifying highs</p><p>early in trend development, so the question you need to ask is whether price</p><p>action has the ability to continue. You find the answer by analysing</p><p>surrounding price action.</p><p>From my perspective, I do not believe that the break above the indicator on</p><p>12 August 2005 was a false signal — quite the contrary. Consider the</p><p>following:</p><p>If you identified the first close above the initial high indicator (on 12</p><p>August 2005), you would have no idea in advance that a dip in price was</p><p>about to follow, so the signal was genuine.</p><p>If the exploration process identified the second close above price action</p><p>(on 2 September 2005), you could see this as continued bullish activity.</p><p>The concept is not just about identifying a singular point and a somewhat</p><p>mechanical point of entry, although it can be</p><p>the old-fashioned</p><p>way using paper trading in real time or by using a simulation program such as</p><p>TradeSim. You should never apply a strategy that has not undergone rigorous</p><p>assessment.)</p><p>Under normal circumstances, I avoid providing excessive detail on</p><p>individual programs; however, TradeSim is used extensively throughout this</p><p>book in an attempt to quantify the effectiveness — or, in some case, the</p><p>incompetence — of different techniques and practices. With this in mind, I</p><p>considered you needed to have some insight as to how such results were</p><p>attained and the validity of the results shown. I also felt that you needed to</p><p>have a basic understanding of how to interpret the information included in the</p><p>results. Many of the values are fairly seIf-explanatory, and these will not be</p><p>cOvered; however, some of the more important aspects of the development</p><p>process will receive attention where necessary.</p><p>Testing parameters and analysis process</p><p>As mentioned, I used TradeSim to simulate the trading process for various</p><p>techniques and processes, and so compare the effectiveness of each.</p><p>TradeSim comes with a 250-page instruction manual that is a book in itself,</p><p>so, as you can appreciate, the list of features is comprehensive. As it is</p><p>impossible to cOver all of the features in TradeSim and do the program</p><p>justice, I will limit the discussion to the basics of getting started. While</p><p>TradeSim is designed to test at the most intricate levels, it is not essential to</p><p>know the inner workings of all the available features in order to produce</p><p>reliable results relative to how we intend to approach the market.</p><p>For publication purposes, I have condensed the reports by removing any</p><p>areas of duplication; however, none of the values has been altered in any</p><p>way. The actual results you see are as produced by TradeSim. While I am</p><p>unable to guarantee the accuracy of the results produced by TradeSim, I have</p><p>no reason to question the validity of the simulation process or the results</p><p>produced. During my extensive use of the program Over years of application,</p><p>I have seldom experienced results that led me to question the integrity of the</p><p>software. At the end of each chapter, I have included the necessary coding</p><p>that will allow you to independently test the results for yourself.</p><p>For testing purposes, including all discussions from this point, I have</p><p>applied the following parameters:</p><p>all strategies are tested on the top 300 stocks listed on the ASX</p><p>testing commenced on 1 January 2000 and concluded on 30 June 2005</p><p>I have loaded 350 weeks or 1500 days, depending on the time frame</p><p>being assessed.</p><p>I arrived at the results included in this book by, firstly, evaluating each step</p><p>of the development process using Monte Carlo analysis. I then located an</p><p>individual simulation that was consistent with the average results produced</p><p>by the Monte Carlo process.</p><p>Monte Carlo analysis</p><p>Before you start scratching your head about Monte Carlo analysis, you first</p><p>need to understand how TradeSim operates. TradeSim trades in the identical</p><p>manner to real-time application in that, once full capital allocation has been</p><p>achieved, all further trades are ignored until an existing position is closed.</p><p>This is then carried through the entire testing period, whereby a new position</p><p>is opened only when trading capital becomes available. One of the more</p><p>obvious problems with some older testing programs is that they use the same</p><p>starting point; therefore, the same sequence of trades unfolds leading to the</p><p>same inevitable outcome. This is a random and single snapshot of many</p><p>potential outcomes. As this is a one-off example of the strategy at work, there</p><p>is the high probability that the result is inconsistent with average strategy</p><p>performance. In order to evaluate the true effectiveness of any strategy, many</p><p>simulations must be run, including a random and an indiscriminate sequence</p><p>of trades during the testing period. Only by exploring many combinations and</p><p>trading sequences can you actually begin to gauge a strategy’s true</p><p>effectiveness.</p><p>This is where Monte Carlo analysis comes into its own. In a conventional</p><p>testing program, the start is always point ‘A’. The same sequence of trades is</p><p>then followed, always using path ‘A’, thereby producing the same result. The</p><p>Monte Carlo process will start at point A and then follow a sequence of</p><p>trades using path A, trading the portfolio in a normal manner, opening and</p><p>closing positions as entry and exit signals occur assuming capital is available.</p><p>The Monte Carlo testing program then goes back and starts the process again,</p><p>but this time it starts at point A and follows a new sequence of trades using</p><p>path B. Once complete, it goes back and starts again — but this time it starts</p><p>at point B, following a new sequence of trades using path C, and so on.</p><p>I think you can begin to understand that Monte Carlo analysis allows you to</p><p>assess the effectiveness of your strategy by evaluating many individual</p><p>simulations that have their own unique sequence of trades independent of all</p><p>other simulations. By reckoning the results, you can obtain a more reliable</p><p>picture of how the technique will perform.</p><p>Using 2000 separate simulations, I performed Monte Carlo analysis for</p><p>every test in order to define an average result; however, TradeSim will allow</p><p>for 20 000 individual simulations if needed. There is no need to become</p><p>embroiled with Monte Carlo analysis or be Overly analytical, as I am simply</p><p>looking to establish a basic benchmark in order to evaluate general</p><p>improvements. Therefore, I will stick with a fairly basic application. You can</p><p>perform your own in-depth analysis if required.</p><p>In terms of the way I have applied Monte Carlo analysis, it is important to</p><p>note the following:</p><p>Once I identified average performance, I then located an individual</p><p>simulation that was consistent with the system average. These are the</p><p>results shown throughout this book.</p><p>As the testing results shown form only one of many possible outcomes,</p><p>they should not be considered as cast in stone or reflective of an actual</p><p>performance you may trade in real time. Many traders will successfully</p><p>outperform the system average while others will fall well short.</p><p>TradeSim cannot incorporate individual experience, understanding,</p><p>discipline or tolerance levels, all of which directly affect trading results,</p><p>nor can it take future market activity into consideration. As good as</p><p>TradeSim may be, like traders, it can only reference and evaluate</p><p>historical data.</p><p>Figure 1.1 shows the Monte Carlo simulation in progress.</p><p>Figure 1.1: Monte Carlo simulation</p><p>Prior to complex simulation programs, testing trading strategies was a very</p><p>laborious process and, hence, usually Overlooked by many. However, there is</p><p>no excuse today should you develop a desire to become creative in your</p><p>approach.</p><p>Simulation example</p><p>Rather than waffle on for pages about the trade simulation process, I will</p><p>design and test a very basic technique, then work through the testing process.</p><p>This will give newcomers to TradeSim the ability to follow the process, and</p><p>will also highlight the factors that I will focus on when comparing the test</p><p>results of each new concept or strategy through the course of this book.</p><p>For the basic trading technique, let’s say that you are a new trader and that</p><p>you have thought of a brilliant concept. For a trade to be triggered, you</p><p>require the closing price to occur above a short-term moving average. (You</p><p>think this is sheer brilliance and you are sure that nobody has ever considered</p><p>such a shrewd concept.) The icing on the cake is that you also require a</p><p>minimum price change and for this to happen above a long-term moving</p><p>average. Just brilliant!! Finally, the exit has you completely stumped but as</p><p>like most people new to trading you consider the exit to be unimportant, a</p><p>close below the short-term moving average is chosen as the perfect exit point.</p><p>Hence, an entry trigger occurs when the closing price:</p><p>is above the 21-day moving average for the first time</p><p>has finished 5 per cent higher</p><p>applied in this manner. The</p><p>concept as I intended involves the identification of new highs very early in</p><p>rally or trend development. The key in successful application is looking</p><p>beyond the trigger and understanding the concept. Identifying an initial high</p><p>is the start of the analysis process, not the conclusion.</p><p>Initial high parameters</p><p>The previous chapter focused on applying relative confirmation indicator</p><p>values; however, the confirmation process is only one step in the analysis</p><p>progression. What is the point of applying a price channel that is nicely in</p><p>tune with trend development if the entry signal is as potentially out of balance</p><p>as the confirmation indicators are at 14 periods? Within the process of</p><p>developing the initial high strategy, I wanted to know how it performed in a</p><p>range of market conditions and at a range of parameters. What was the ideal</p><p>number of periods for the initial high strategy and could I apply a generic</p><p>value in both bullish and bearish conditions? While I will not bore you with</p><p>all the details of the process that took many months to compile in real time,</p><p>the main point is that I set about testing the initial high in both a bearish year,</p><p>2002, and then at the start of a solid bull run, in 2003. This was an ideal</p><p>testing ground, as the conditions occurred in the same economic climate with</p><p>the same market participants.</p><p>Figure 8.6 shows the results I achieved through testing the initial high</p><p>technique at different time periods in 2002.</p><p>Figure 8.6: initial high performance in 2002 testing various periods</p><p>My conclusions tended to contradict what I had been repeatedly told and</p><p>what is accepted market practice. In bearish conditions traders are told to</p><p>shorten their strategies, and so enter and exit quickly to capture rallies before</p><p>they mature or collapse. This may be a logical approach for a rally trader but</p><p>can be a complete disaster for anyone still interested in position trading. My</p><p>testing showed that shortening the initial high signal meant results</p><p>deteriorated noticeably. It was not until the initial high was extended out to</p><p>17-day highs that the results began to improve. Keep in mind how the</p><p>concept works — a longer period indicator remains sufficiently above down-</p><p>trending price action, absorbing some rally activity without generating entry</p><p>signals. This is shown in figure 8.7.</p><p>Figure 8.7: entry techniques and trend development</p><p>As figure 8.6 shows, if initial highs were traded up to 17 periods, a return in</p><p>the range of 15 per cent to 23 per cent could have been expected during 2002.</p><p>Once the initial high indicator was extended to between 19 and 23 periods,</p><p>the return began to improve markedly, peaking at 43 per cent. This is not a</p><p>bad return considering that the market in general declined by 12.25 per cent.</p><p>Extending the initial high periods beyond 23 periods failed to produce the</p><p>same positive result, as the annualised return declined exponentially as prices</p><p>were extended. This decline in annualised return most likely stemmed from</p><p>an excessively delayed entry.</p><p>Position traders need to identify the point where minor rallies are absorbed.</p><p>As shown in figure 8.6, using an initial high ranging from 19 to 23 periods in</p><p>bearish conditions produced the highest region of probability, as minor and</p><p>insignificant rallies were separated from more legitimate trending activity.</p><p>Beyond 23 periods, the signal is becoming too slow and is starting to eat into</p><p>profits by excessively delaying the entry.</p><p>These results show that for position traders it is best to lengthen entry</p><p>signals in bearish conditions, rather than shorten them as many suggest.</p><p>In 2003 the opposite market conditions to 2002 were produced, whereby</p><p>the market experienced a solid rebound from the March low. As the market</p><p>behaved in the opposite manner, so did the initial high and a longer initial</p><p>high indicator actually impeded returns.</p><p>Figure 8.8 shows the results achieved through testing the initial high</p><p>technique in 2003 at different periods.</p><p>Figure 8.8: initial high performance in 2003 testing various periods</p><p>As figure 8.8 shows, if a very short initial high indicator of between 7 and 13</p><p>periods had been applied in 2003, in theory you could have posted some</p><p>pretty impressive results. While it’s unlikely that you would have actually</p><p>achieved this — if for no other reason than the massive number of signals</p><p>produced by a very short initial high indicator — in theory, it was there for</p><p>the taking.</p><p>Between 19 and 25 periods the initial high indicator produced its worst</p><p>result, with returns between 30 per cent and 40 per cent. Even the greatly</p><p>reduced return of 40 per cent is not too shabby.</p><p>From a 25-period initial high, results steadily improved; however, beyond</p><p>35 periods, additional gains were minimal.</p><p>These results show that in bullish conditions it is best to shorten your entry</p><p>signal if you are going to trade at optimum efficiency. Again this is contrary</p><p>to what many suggest; however, I believe that many confuse the role of entry</p><p>triggers with exit strategies. Lengthening or shortening your entry technique</p><p>has absolutely nothing to do with determining how long you remain with</p><p>trending price. The research above tends to suggest the exact opposite to</p><p>accepted practice, whereby you should lengthen your signal in bearish</p><p>conditions while shortening the indicator length in a bullish climate.</p><p>As figure 8.6 and 8.8 show, I also tested an initial 250-day high during</p><p>2002 and 2003. While trading yearly highs may be a time-proven method,</p><p>this approach produced the worst results in 2002 at 250 days, with a net</p><p>return just above break-even. Trading initial 250-day highs in 2003 was a</p><p>completely different matter. This would, in theory, have posted a return of</p><p>around 80 per cent, on the back of surprisingly high consistency.</p><p>Now you need to ask yourself the following question, and you need to be</p><p>honest — am I trading on values that are relative to my strategy, or am I</p><p>simply using values that everyone uses or, worse still, alternative values for</p><p>the sake of being different?</p><p>Some inner honesty here may highlight why things are not going so well.</p><p>Establishing a benchmark</p><p>Having defined a suitable entry strategy, I will now produce a benchmark for</p><p>comparison. If you are set in your ways and not open to change or prepared</p><p>to have existing beliefs challenged, it’s probably time to skip to the next</p><p>chapter. If you genuinely believe that no better indicator exists than a 14-</p><p>period relative strength index, rest assured — you are not going to like what</p><p>you are about to read.</p><p>Testing parameters</p><p>The strategy has developed to the point where I will apply the following</p><p>parameters:</p><p>initial high periods — 21</p><p>time frame — daily</p><p>entry signal — a close above initial high indicator</p><p>confirmation — no means of confirmation has been applied</p><p>exit strategy — 2.7 × relative percentage trailing stop with a close</p><p>below.</p><p>Again I have not Overly complicated the trading process. In order to</p><p>determine whether the approach to relative analysis is as the name suggests, I</p><p>have first tested the initial high strategy in its raw form with no confirmation</p><p>in the way of price channels, volume support or price change. This is not</p><p>exactly how I apply the initial high technique, but to either prove or disprove</p><p>the concept of relative analysis, a benchmark that represents the entry signal</p><p>in its purest form is needed to begin with. I have continued on a daily time</p><p>frame in order to show that I am not specifically targeting one element of</p><p>price action in order to favour a particular result.</p><p>Test results</p><p>The test results for the initial high benchmark, also shown in table 8.1, are as</p><p>follows:</p><p>Table 8.1: initial high — benchmark</p><p>Profit summary</p><p>Profit status PROFITABLE</p><p>Starting capital $25 000.00</p><p>Finishing capital $43 359.04</p><p>Maximum equity/(date) $25 206.11 (25/02/2005)</p><p>Minimum equity/(date) −$3212.37(20/06/2000)</p><p>Gross trade profit $62 250.06 (249.00%)</p><p>Gross trade loss −$43 891.02 (−175.56%)</p><p>Total net profit $18 359.04(73.44%)</p><p>Average profit</p><p>per trade $108.63</p><p>Profit factor 1.4183</p><p>Profit index 29.49%</p><p>Total transaction cost $6760.00</p><p>Total slippage $0.00</p><p>Daily compound interest rate 0.0263%</p><p>Annualised compound interest rate 10.0637%</p><p>Trades processed 3630</p><p>Trades taken 169</p><p>Partial trades taken 0</p><p>Trades rejected 3461</p><p>Winning trades 70 (41.42%)</p><p>Losing trades 99 (58.58%)</p><p>Largest winning trade/(date) $8096.89 (21/02/2001)</p><p>Largest losing trade/(date) −$1 799.02 (12/11/2001)</p><p>Average winning trade $889.29</p><p>Average losing trade −$443.34</p><p>Average win/average loss 2.0059</p><p>Trade duration statistics (all trades)</p><p>Maximum trade duration 353 (days)</p><p>Minimum trade duration 3 (days)</p><p>Average trade duration 74 (days)</p><p>Winning trades</p><p>Maximum trade duration 353 (days)</p><p>Minimum trade duration 20 (days)</p><p>Average trade duration 117 (days)</p><p>Losing trades</p><p>Maximum trade duration 174 (days)</p><p>Minimum trade duration 3 (days)</p><p>Average trade duration 43 (days)</p><p>Consecutive trade statistics</p><p>Maximum consecutive winning trades 8</p><p>Maximum consecutive losing trades 11</p><p>Average consecutive winning trades 2.00</p><p>Average consecutive losing trades 2.75</p><p>Relative drawdown</p><p>Maximum dollar drawdown/(date) $5275.02 (19/04/2005)</p><p>Maximum percentage drawdown/(date) 11.3300% (20/06/2000)</p><p>Absolute (peak-to-valley) dollar drawdown</p><p>Maximum dollar drawdown $7793.40 (15.5700%)</p><p>Capital peak/(date) $50 046.11 (25/02/2005)</p><p>Capital valley/(date) $42 252.72 (31/08/2005)</p><p>The initial high entry trigger on its own applied on a daily time frame</p><p>proved profitable 41.42 per cent of the time. The average profit was</p><p>$889.29 while the average loss was very close to half at -$443.34, thereby</p><p>producing a win/loss ratio of 2:1.</p><p>The peak to valley drawdown was 15.57 per cent so the worse-case</p><p>scenario from capital high to low is not Overly pleasing; however, it is</p><p>certainly not extreme for a daily strategy. I have seen far worse peak to</p><p>valley drawdowns than that produced by the benchmark here.</p><p>The number of potential trading opportunities is completely</p><p>unacceptable at 3461. Remember — no means for filtering the raw signal</p><p>has been applied, so a reasonably short-term entry trigger would be</p><p>expected to produce a relatively high number of signals.</p><p>The total net profit was $18 359.04 or 73.44 per cent, which equates to</p><p>an annualised return of 10.06 per cent. While the raw signal tested</p><p>profitable, I won’t retire at this rate.</p><p>Equity curve and yearly profits</p><p>As shown in figure 8.9, the equity curve for the benchmark was less than</p><p>ideal, especially from late February 2005 where it was on the downhill run.</p><p>The equity curve shows that the strategy tended to follow market behaviour,</p><p>where strong bullish conditions saw a rise in the equity curve and non-</p><p>trending or bearish behaviour in the equity curve coincided with the</p><p>corresponding market direction. Such behaviour would be expected,</p><p>considering that no filtering mechanisms were used to help screen weak or</p><p>less than ideal trading opportunities.</p><p>Figure 8.9: equity curve for initial high benchmark</p><p>Figure 8.10 shows the yearly profits for the initial high benchmark.</p><p>Figure 8.10: yearly profits for initial high benchmark</p><p>The yearly profit graph is reflective of general market behaviour. The</p><p>benchmark experienced a loss in 2002, which is consistent with the negative</p><p>year experienced by the broader market, while 2003–2004 produced positive</p><p>results in unison with a bullish trading environment.</p><p>Beyond the benchmark — adding</p><p>confirmation</p><p>The first step in moving beyond the benchmark and so improving results is to</p><p>apply the initial high strategy in conjunction with a 14-period relative price</p><p>channel. This is reflective of a 14-period relative strength index. This time an</p><p>entry signal generated by the initial high indicator — that is, where the</p><p>closing price finishes above the initial high indicator — must also occur</p><p>within the bullish band of the relative price channel. This effectively confirms</p><p>that the initial high entry signal is legitimate, or at least enjoys a high degree</p><p>of probability. I wish to advise that the following results may offend some</p><p>readers.</p><p>Testing parameters</p><p>initial high periods — 21</p><p>time frame — daily</p><p>entry signal — a close above the initial high indicator</p><p>confirmation — 14-period relative price channel</p><p>exit strategy — 2.7 x relative percentage trailing stop with a close</p><p>below.</p><p>Test results</p><p>So what impact did the relative price channel of 14 periods have on initial</p><p>high efficiency? Absolutely none.</p><p>The test results, also shown in table 8.2, were as follows:</p><p>Table 8.2: initial high with 14-period relative price channel as</p><p>confirmation</p><p>Profit summary</p><p>Profit status PROFITABLE</p><p>Starting capital $25 000.00</p><p>Finishing capital $43 538.45</p><p>Maximum equity/(date) $27 288.58 (8/03/2005)</p><p>Minimum equity/(date) −$2493.10 (8/06/2000)</p><p>Gross trade profit $69 595.67 (278.38%)</p><p>Gross trade loss −$51 057.22 (−204.23%)</p><p>Total net profit $18 538.45 (74.15%)</p><p>Average profit per trade $104.74</p><p>Profit factor 1.3631</p><p>Profit index 26.64%</p><p>Total transaction cost $7080.00</p><p>Total slippage $0.00</p><p>Daily compound interest rate 0.0264%</p><p>Annualised compound interest rate 10.1074%</p><p>Trades processed 3510</p><p>Trades taken 177</p><p>Partial trades taken 0</p><p>Trades rejected 3333</p><p>Winning trades 74 (41.81%)</p><p>Losing trades 103 (58.19%)</p><p>Largest winning trade/(date) $7258.63 (31/07/2001)</p><p>Largest losing trade/(date) −$1831.38 (8/08/2002)</p><p>Average winning trade $940.48</p><p>Average losing trade −$495.70</p><p>Average win/average loss 1.8973</p><p>Trade duration statistics (all trades)</p><p>Maximum trade duration 311 (days)</p><p>Minimum trade duration 3 (days)</p><p>Average trade duration 75 (days)</p><p>Winning trades</p><p>Maximum trade duration 311 (days)</p><p>Minimum trade duration 8 (days)</p><p>Average trade duration 124 (days)</p><p>Losing trades</p><p>Maximum trade duration 173 (days)</p><p>Minimum trade duration 3 (days)</p><p>Average trade duration 41 (days)</p><p>Consecutive trade statistics</p><p>Maximum consecutive winning trades 7</p><p>Maximum consecutive losing trades 11</p><p>Average consecutive winning trades 1.95</p><p>Average consecutive losing trades 2.71</p><p>Relative drawdown</p><p>Maximum dollar drawdown/(date) $5 520.18 (24/06/2005)</p><p>Maximum percentage drawdown/(date) 11.9200% (8/06/2000)</p><p>Absolute (peak to valley) dollar drawdown</p><p>Maximum dollar drawdown $9908.39 (21.6000%)</p><p>Capital peak/(date) $45 880.02 (17/05/2002)</p><p>Capital valley/(date) $35 971.64 (4/02/2004)</p><p>The rate of profitable trades lifted by a staggering amount — from</p><p>41.42 per cent to 41.81 per cent. However, the average loss actually</p><p>increased rather than decreased, which reduced the average win/loss ratio</p><p>to 1.89:1. By incorporating a 14-period RSI as a means of confirmation,</p><p>the strategy has gone backwards, whereby the average loss has</p><p>deteriorated to $495.70 from the original average loss of $443.34.</p><p>While the number of potential trading opportunities did decrease</p><p>slightly, it remains unmanageable at 3510.</p><p>The big loser was the trading capital, which experienced a peak to</p><p>valley drawdown of 21.60 per cent. This is completely unacceptable and</p><p>is worse than the peak to valley drawdown of 15.57 per cent achieved by</p><p>the benchmark.</p><p>The annualised return was a marginal improvement at best with a return</p><p>of 10.10 per cent, which is still disappointing.</p><p>The relative price channel at 14 periods is a complete waste of time when</p><p>used in conjunction with the initial high technique — as the test results show,</p><p>the strategy actually went backwards. You would have done just as well</p><p>applying no confirmation analysis.</p><p>I have repeatedly suggested that indicators when applied on a daily time</p><p>frame are usually nothing more than confidence triggers, not confirmation. At</p><p>the time, I did not have the benefit of TradeSim, just experience that led me</p><p>to draw such conclusions. Not only has TradeSim given me the ability to</p><p>confirm what I have suspected for a long time, but it has allowed me to</p><p>actually quantify this largely ineffective relationship.</p><p>Equity curve and yearly profits</p><p>Figure 8.11 shows the equity curve for the initial high strategy with 14-period</p><p>relative strength channel as confirmation.</p><p>Figure 8.11: equity curve for initial high strategy using</p><p>14-period relative</p><p>price channel</p><p>This equity curve is a poor ring-in for what should constitute an equity curve.</p><p>It’s all Over the place and seldom correlates with general market behaviour.</p><p>There was a solid rise from late 2001 into 2002, but from mid-2002 through</p><p>to early 2004, everything was downhill. The decline was constant Over an</p><p>18-month period, which would have been disappointing given the solid bull</p><p>run of 2003.</p><p>Figure 8.12 shows the yearly profits for the initial high strategy using the</p><p>14-period relative strength channel as confirmation. The inconsistent nature</p><p>of confirmation is more readily apparent on the yearly profit graph. In 2002</p><p>the broader market declined yet the strategy produced its best result. This was</p><p>immediately followed, however, by a solid loss in 2003 that coincided with</p><p>the early stages of a strong bull run.</p><p>Figure 8.12: yearly profits for initial high 14-period relative price</p><p>channel</p><p>The strategy lost money in a bullish environment but produced a positive</p><p>result in a bearish market. Are you starting to see why those who rely heavily</p><p>on indicators — especially indicators that are not relative to general trend</p><p>development — really struggle to survive? The problem is not always or</p><p>solely the entry trigger or the trailing stop, but ineffective confirmation tools.</p><p>Tip</p><p>When a strategy fails, the first two elements in line for review are the entry trigger and the</p><p>trailing stop. The confirmation tool seldom receives in-depth scrutiny due to the perceived</p><p>relationship that exists between the indicator and trend development. Traders assume that a</p><p>legitimate relationship exists for no other reason than that on the surface there appears to be</p><p>a degree of correlation. Prominent points in price action and indicator development appear</p><p>to occur simultaneously. What traders fail to consider is that prominent points in price</p><p>action and indicator development usually coincide with both short and longer term</p><p>indicator application.</p><p>Beyond the benchmark — 55-period</p><p>relative channel as confirmation</p><p>Once again, the price channel is used as confirmation for an entry signal —</p><p>that is, the signal must occur within the bullish band of the price channel.</p><p>This time around I have extended the parameters out to 55 periods, which</p><p>creates a very slow relative price channel. Many would suggest that 55</p><p>periods is far too slow; however, the default values have hardly worked with</p><p>any conviction, so what is to be lost by applying values that are more relative</p><p>to price action?</p><p>Testing parameters</p><p>The following testing parameters were applied:</p><p>initial high periods — 21</p><p>time frame — daily</p><p>entry signal — a close above the initial high indicator</p><p>confirmation — 55-period relative strength channel</p><p>exit strategy — 2.7 x relative percentage trailing stop.</p><p>Test results</p><p>Regardless of any preconceived notions you may have about a slow price</p><p>channel, the testing results cannot be argued with. The results, also shown in</p><p>table 8.3, were as follows:</p><p>Table 8.3: initial high with 55-period relative price channel as</p><p>confirmation</p><p>Profit summary</p><p>Profit status PROFITABLE</p><p>Starting capital $25 000.00</p><p>Finishing capital $94 079.19</p><p>Maximum equity/(date) $69 079.19(7/10/2005)</p><p>Minimum equity/(date) −$4657.40 (18/04/2000)</p><p>Gross trade profit $113 229.83 (452.92%)</p><p>Gross trade loss −$44 150.64 (−176.60%)</p><p>Total net profit $69 079.19 (276.32%)</p><p>Average profit per trade $314.00</p><p>Profit factor 2.5646</p><p>Profit index 61.01%</p><p>Total transaction cost $8800.00</p><p>Total slippage $0.00</p><p>Daily compound interest rate 0.0631%</p><p>Annualised compound interest rate 25.9032%</p><p>Trades processed 1201</p><p>Trades taken 220</p><p>Partial trades taken 0</p><p>Trades rejected 981</p><p>Winning trades 105 (47.73%)</p><p>Losing trades 115 (52.27%)</p><p>Largest winning trade/(date) $11 508.15 (26/03/2001)</p><p>Largest losing trade/(date) −$1,952.94 (18/04/2000)</p><p>Average winning trade $1 078.38</p><p>Average losing trade −$383.92</p><p>Average win/average loss 2.8089</p><p>Trade duration statistics (all trades)</p><p>Maximum trade duration 425 (days)</p><p>Minimum trade duration 1 (days)</p><p>Average trade duration 79 (days)</p><p>Winning trades</p><p>Maximum trade duration 425 (days)</p><p>Minimum trade duration 5 (days)</p><p>Average trade duration 114 (days)</p><p>Losing trades</p><p>Maximum trade duration 136 (days)</p><p>Minimum trade duration 1 (days)</p><p>Average trade duration 47 (days)</p><p>Consecutive trade statistics</p><p>Maximum consecutive winning trades 6</p><p>Maximum consecutive losing trades 7</p><p>Average consecutive winning trades 1.94</p><p>Average consecutive losing trades 2.13</p><p>Relative drawdown</p><p>Maximum dollar drawdown/(date) $4117.93 (18/04/2000)</p><p>Maximum percentage drawdown/(date) 16.8500% (18/04/2000)</p><p>Absolute (peak to valley) dollar drawdown</p><p>Maximum dollar drawdown $6071.41 (13.4400%)</p><p>Capital peak/(date) $45 159.20 (29/04/2002)</p><p>Capital valley/(date) $39 087.80 (17/03/2003)</p><p>The number of profitable trades has risen to 47.73 per cent, up from</p><p>41.81 per cent using the 14-period relative strength channel. While this</p><p>may not appear significant on the surface, this is a 14.16 per cent</p><p>improvement in regard to the level of consistency. Are you still denying</p><p>the obvious?</p><p>The average loss declined from $495.70 to $383.92. Still in denial?</p><p>The average profit, at $1078.38, is a marginal improvement Over the</p><p>14-period relative strength channel, which returned an average result of</p><p>$940.48.</p><p>The annualised return jumped from a very conservative 10.11 per cent</p><p>to a more respectable 25.90 per cent. Again, this may not appear</p><p>significant at first, but it is a 156.43 per cent improvement Over the 14-</p><p>period relative price channel. How are you going with all of this? And,</p><p>no, you cannot close the book or skip to the next chapter. Still shaking</p><p>your head with disapproval? Keep shaking.</p><p>The peak to valley drawdown was also a marked improvement. The 14-</p><p>period relative price channel produced a maximum drawdown of 21.60</p><p>per cent. The 55-period channel almost halved this negative aspect by</p><p>producing a drawdown of 13.44 per cent, which tends to suggest that the</p><p>big losers have been avoided.</p><p>Equity curve and yearly profits</p><p>Figure 8.13 shows the equity curve for the initial high strategy using a 55-</p><p>period relative strength channel as confirmation.</p><p>Figure 8.13: equity curve for initial high strategy using 55-period relative</p><p>price channel</p><p>The equity curve for the 55-period relative price channel is a vast</p><p>improvement Over previous attempts. A profitable outcome has been</p><p>generated while an increased level of stability has been achieved. The loss of</p><p>$158.55 incurred during 2002 is to be expected, and generally the level of</p><p>consistency is fairly impressive — especially considering that no other means</p><p>of confirmation or filtering, such as volume or price change or long-term</p><p>moving average, have been applied.</p><p>Looking at the yearly profits chart, shown in figure 8.14, the degree of</p><p>consistency is again relatively impressive. The only exception was 2000,</p><p>which performed well below par. However, the test results still show that this</p><p>is a confirmation technique that can filter signals under a range of conditions.</p><p>How are you handling the concept of applying relative values? The testing</p><p>in this chapter commenced with an initial high entry trigger with no</p><p>confirmation. Although the technique produced a positive result, it was</p><p>somewhat conservative and erratic. A 14-period relative price channel, which</p><p>is effectively a 14-period relative strength index, was then incorporated and</p><p>the results actually went backwards. While this will come as a shock to some,</p><p>it came as no surprise to me.</p><p>Figure 8.14: yearly profits for initial high strategy using 55-period</p><p>relative price channel</p><p>The approach was finally made more relative as general trend development</p><p>was allowed to determine the parameters that were to be applied with the</p><p>relative price channel. The results improved dramatically for all aspects</p><p>involving price channel application. I know some of you will be clinging to</p><p>your old ways. The simple fact is the default values have not worked as</p><p>expected when combined with the initial high technique.</p><p>The end</p><p>result</p><p>So many times a really solid strategy can be defined through research and</p><p>testing but, when it comes to real-time application, it’s a complete nightmare.</p><p>Having revised the channel parameters in order to reflect general trend</p><p>development, what does the end result look like? As you can see in figure</p><p>8.15, price action, the entry trigger and how it relates to the price channel are</p><p>easily discernible from a visual perspective.</p><p>Figure 8.15: entry signals from initial high strategy using 55-period</p><p>relative price channel</p><p>As can be seen in figure 8.15, Australian Stock Exchange (ASX) has three</p><p>legitimate entry signals as tested. There also seems to be one questionable</p><p>signal — or is it?</p><p>Let’s first look at the fist signal. Would you enter on this entry trigger?</p><p>While the entry trigger is genuine, it has developed relatively close to the</p><p>upper cord of the bullish band, so you could reasonably assume that the</p><p>potential upside Over the short term may be limited. While this may or may</p><p>not later prove to be the case, probability dictates that the odds favour a pull</p><p>back or a pause in the not too distant future, so you should seriously consider</p><p>holding off on an entry.</p><p>If there happened to be the perfect trigger, signal two is it. The closing</p><p>price finished in the bullish band for the first time while also breaking above</p><p>the initial high indicator. The stock gapped up while closing near the high.</p><p>The break to the upside did not develop from rebounding price action but</p><p>from consolidating price action that developed in the middle region of the</p><p>neutral zone. Signals don’t come much better than this.</p><p>An additional signal, signal three, occurred following a bullish period of</p><p>consolidation. As price action is well down from the upper cord of the bullish</p><p>band, there is sufficient room for ongoing upside. While the majority of</p><p>initial high triggers seldom develop with the clinical precision of signal two,</p><p>providing the signal occurs some distance from the upper cord, there is still</p><p>sufficient room for potential upside before a probable retracement.</p><p>What about the questionable signal? How should you deal with a signal</p><p>such as this? Price action has consolidated in the neutral zone without</p><p>rebounding from either extreme bullish or bearish values, so this is</p><p>consolidation in the true sense of the word. Price action has moved above the</p><p>initial high entry trigger yet remains in the neutral zone. As it is not on the</p><p>rebound, this is a legitimate entry signal — but more on this in the next</p><p>chapter.</p><p>Using the lessons so far</p><p>How does the concept of relative analysis relate to your approach? Are you</p><p>beginning to question popular parameters and general methods of indicator</p><p>application, especially for position trading? The question you should be</p><p>starting to ask yourself now is whether complex techniques and methods of</p><p>confirmation are more important than relative techniques. So many</p><p>inexperienced traders believe that increased complexity and additional</p><p>confirmation tools will somehow restore balance to an out of balance</p><p>strategy. This is not the case. The key is not, and never has been, added</p><p>complexity — the key is understanding the relationship between price and</p><p>indicator.</p><p>Generally speaking, the more responsive an indicator becomes, the less</p><p>consistent the outcome will be — if for no other reason than it identifies</p><p>every bump in price action.</p><p>It may be far easier for some to simply dismiss what has been discussed so</p><p>far as not appropriate to you, as you use a different entry trigger and so can</p><p>continue on as usual. To re-evaluate existing techniques at such a</p><p>fundamental level only to discOver that what they have been applying all this</p><p>time has been largely ineffective will be too much for some people to accept,</p><p>as it will question the very foundation that identifies them as a trader. Having</p><p>to admit to an error in understanding may not sit well with the ego. If you feel</p><p>like this, don’t worry — you are not the odd one out. I travelled this same</p><p>path as I unearthed the relative adaptation of indicators to price action. I</p><p>swallowed the same bitter pill — thankfully, it only has to be swallowed</p><p>once.</p><p>I will now move onto testing the initial high strategy combined with a</p><p>different confirmation channel and at different periods.</p><p>Initial high and daily comparative strength</p><p>channel</p><p>I combined the initial high strategy with the 34-period comparative strength</p><p>channel Over a daily time frame.</p><p>Testing parameters</p><p>The testing parameters for this strategy were as follows:</p><p>initial high periods — 21</p><p>entry trigger — a close above the initial high; the closing price must be</p><p>in the bullish band at the time</p><p>exit trigger — a close below the 2.7× relative percentage trailing stop.</p><p>Test results</p><p>Combining the initial high strategy with the daily comparative strength</p><p>channel produced the following results, also shown in table 8.4:</p><p>Table 8.4: initial high strategy using daily 34-period comparative</p><p>strength channel</p><p>Profit summary</p><p>Profit status PROFITABLE</p><p>Starting capital $25 000.00</p><p>Finishing capital $96 550.82</p><p>Maximum equity/(date) $71 550.82 (14/10/2005)</p><p>Minimum equity/(date) −$1964.61 (14/04/2000)</p><p>Gross trade profit $118 437.36 (473.75%)</p><p>Gross trade loss −$46 886.54 (−187.55%)</p><p>Total net profit $71 550.82 (286.20%)</p><p>Average profit per trade $304.47</p><p>Profit factor 2.5260</p><p>Profit index 60.41%</p><p>Total transaction cost $9400.00</p><p>Total slippage $0.00</p><p>Daily compound interest rate 0.0641%</p><p>Annualised compound interest rate 26.3733%</p><p>Trades processed 914</p><p>Trades taken 235</p><p>Partial trades taken 0</p><p>Trades rejected 679</p><p>Winning trades 111 (47.23%)</p><p>Losing trades 124 (52.77%)</p><p>Largest winning trade/(date) $11 508.15 (26/03/2001)</p><p>Largest losing trade/(date) −$1952.94 (18/04/2000)</p><p>Average winning trade $1067.00</p><p>Average losing trade −$378.12</p><p>Average win/average loss 2.8219</p><p>Trade duration statistics (all trades)</p><p>Maximum trade duration 326 (days)</p><p>Minimum trade duration 1 (days)</p><p>Average trade duration 83 (days)</p><p>Winning trades</p><p>Maximum trade duration 326 (days)</p><p>Minimum trade duration 32 (days)</p><p>Average trade duration 127 (days)</p><p>Losing trades</p><p>Maximum trade duration 189 (days)</p><p>Minimum trade duration 1 (days)</p><p>Average trade duration 45 (days)</p><p>Consecutive trade statistics</p><p>Maximum consecutive winning trades 7</p><p>Maximum consecutive losing trades 11</p><p>Average consecutive winning trades 2.09</p><p>Average consecutive losing trades 2.34</p><p>Relative drawdown</p><p>Maximum dollar drawdown/(date) $4175.31 (29/08/2002)</p><p>Maximum percentage drawdown/(date) 8.9030% (18/04/2000)</p><p>An annualised return of 26.73 per cent was produced. This is an</p><p>improvement on the annualised return produced by the initial high</p><p>strategy using a 55-period relative strength channel.</p><p>The average loss declined significantly from the original average loss</p><p>of $1003.77 produced by the comparative strength benchmark in chapter</p><p>5 to a loss of $378.12.</p><p>Unfortunately, the average profit was also restricted to $1067.00. As</p><p>shown above, the relative percentage trailing stop was used for position</p><p>management, rather than a close below the lower cord of the bullish band.</p><p>It would be consistent with upcoming discussions to expect a superior</p><p>result using the bear range or mean close trailing stops; however as this is</p><p>discussed in considerably more detail in chapters 10 and 11, I will not</p><p>pre-empt those discussions here.</p><p>The relative drawdown, at 8.90 per cent, happened to be the same as</p><p>the peak to valley drawdown; this negative impact on capital is more than</p><p>acceptable.</p><p>The number of potential trading opportunities declined from the 1201</p><p>produced by the initial high strategy using a 55-period relative price</p><p>channel to 914, which is a sizeable improvement.</p><p>Some of you may be questioning the validity of such an approach, arguing</p><p>that the improvement is marginal at best. Remember — in this approach, the</p><p>top 300 stocks are being tested. The stocks are then compared against the</p><p>XJO. The bulk of stocks being testing against also make up the XJO, so the</p><p>results are going to be very similar. The application involves the</p><p>identification of those few stocks that change exponentially compared with</p><p>the index. The key to trading on strength is consistency rather than additional</p><p>gain. By applying the identical process using a comparative strength channel</p><p>rather than a relative price channel, 287 potentially false signals were</p><p>removed. Why false? Because while the end result did not change, the</p><p>number of signals produced was slashed and the annualised return increased</p><p>marginally.</p><p>Also remember that I have not spent hours with TradeSim searching for the</p><p>optimum values. This means substantially better results than this could</p><p>almost certainly exist. Not all elements of the trading process revolve around</p><p>the net return. Have you ever wondered about the effect of removing almost</p><p>300 potentially meaningless signals?</p><p>Equity curve and yearly profits</p><p>Figure 8.16 shows the equity curve for the initial high strategy using a daily</p><p>comparative strength channel. As can be seen in figure 8.16, when the initial</p><p>high strategy was applied in conjunction with the comparative strength</p><p>channel, a reasonable degree of consistency was achieved. The equity curve</p><p>indicates solid growth through 2001, followed by a loss of 2.9458 per cent in</p><p>2002. From March 2003, it was reasonably smooth sailing, with consistent</p><p>results leading up to the conclusion of the testing period. The level of</p><p>consistency experienced by the strategy makes it more likely that it can be</p><p>applied in real time. The real question with regard to consistency occurs</p><p>when the initial high technique is applied on weekly data.</p><p>Figure 8.16: equity curve for initial high strategy using daily 34-period</p><p>comparative strength channel</p><p>With the exception of 2002, where the strategy produced a −2.94 per cent</p><p>return, there was a degree of consistency in the returns produced. As shown</p><p>in figure 9.17, the strategy produced returns of 32.95 per cent, 18.26 per cent</p><p>and 19.90 per cent since the bull run commenced in March 2003. Some may</p><p>say that the strategy underperformed the market because of the focus on</p><p>consistency; however, this is a relatively short-sighted perspective. I aim for</p><p>consistency of returns in a range of conditions. Sure, the results could be</p><p>improved with some finetuning, but this may see 2002 produce a negative</p><p>result. Always remember the bigger picture.</p><p>Figure 8.17: yearly profits from initial high strategy using daily 34-</p><p>period comparative strength channel</p><p>Initial high and weekly comparative</p><p>strength channel</p><p>For the final strategy I will look at in this chapter, I again combined the initial</p><p>high strategy with the comparative strength channel, but I changed some of</p><p>the parameters. Instead of a 21-period initial high, I used 17 periods. I again</p><p>used a 34-week comparative strength channel, but I applied it to a weekly,</p><p>rather than a daily, time frame. Making these changes produced results that</p><p>were interesting to say the least.</p><p>Testing parameters</p><p>The testing parameters for this strategy are as follows:</p><p>initial high periods — 17</p><p>entry trigger — a close above the initial high; the closing price must be</p><p>in the bullish band at the time</p><p>exit trigger — a close below the 2.7 × relative percentage trailing stop.</p><p>Test results</p><p>The test results for the strategy, as shown in table 8.5, are as follows:</p><p>Table 8.5: initial high strategy using weekly 34-period comparative</p><p>strength channel</p><p>Profit summary</p><p>Profit status PROFITABLE</p><p>Starting capital $25 000.00</p><p>Finishing capital $115 515.66</p><p>Maximum equity/(date) $91 296.37 (22/04/2005)</p><p>Minimum equity/(date) $2663.40 (12/05/2000)</p><p>Gross trade profit $116 394.73 (465.58%)</p><p>Gross trade loss −$25 879.07 (−103.52%)</p><p>Total net profit $90 515.66 (362.06%)</p><p>Average profit per trade $1005.73</p><p>Profit factor 4.4976</p><p>Profit index 77.77%</p><p>Total transaction cost $3600.00</p><p>Total slippage $0.00</p><p>Daily compound interest rate 0.0727%</p><p>Annualised compound interest rate 30.3609%</p><p>Trades processed 281</p><p>Trades taken 90</p><p>Partial trades taken 0</p><p>Trades rejected 191</p><p>Winning trades 48 (53.33%)</p><p>Losing trades 42 (46.67%)</p><p>Largest winning trade/(date) $14 377.96 (8/03/2002)</p><p>Largest losing trade/(date) −$2039.53 (12/05/2000)</p><p>Average winning trade $2424.89</p><p>Average losing trade −$616.17</p><p>Average win/average loss 3.9354</p><p>Trade duration statistics (all trades)</p><p>Maximum trade duration 777 (days)</p><p>Minimum trade duration 7 (days)</p><p>Average trade duration 191 (days)</p><p>Winning trades</p><p>Maximum trade duration 777 (days)</p><p>Minimum trade duration 7 (days)</p><p>Average trade duration 264 (days)</p><p>Losing trades</p><p>Maximum trade duration 273 (days)</p><p>Minimum trade duration 28 (days)</p><p>Average trade duration 106 (days)</p><p>Consecutive trade statistics</p><p>Maximum consecutive winning trades 8</p><p>Maximum consecutive losing trades 10</p><p>Average consecutive winning trades 2.29</p><p>Average consecutive losing trades 2.10</p><p>Relative drawdown</p><p>Maximum dollar drawdown/(date) $4420.27 (21/02/2003)</p><p>Maximum percentage drawdown/(date) 9.5940% (12/05/2000)</p><p>Absolute (peak to valley) dollar drawdown</p><p>Maximum dollar drawdown $5102.86 (7.7080%)</p><p>Capital peak/(date) $66 203.15 (2/08/2002)</p><p>Capital valley/(date) $61 100.29 (21/03/2003</p><p>The annualised net return is reasonably solid at 30.36 per cent.</p><p>The percentage of winning trades was 53.33 per cent. While this is not</p><p>extreme when compared with other results, very few strategies I have</p><p>encountered manage to produce this kind of success rate with no</p><p>additional means of confirmation. This means that, whenever the closing</p><p>price pushes above the initial 17-week high indicator (providing price</p><p>action was in the bullish band of the comparative strength channel at the</p><p>time), the odds of probability favour a profitable outcome by the time an</p><p>exit signal is produced more often than not. Assuming average profits are</p><p>larger than average losses, there is only one possible outcome.</p><p>The peak to valley drawdown was fairly impressive at 7.70 per cent.</p><p>The average loss, at $616.17, is marginally better than the majority of</p><p>other techniques developed on weekly time frames.</p><p>The average profit came in at $2039.53, producing a win/loss ratio of</p><p>3.93:1, which, again, is more than acceptable for a starting point for</p><p>further refinement.</p><p>The number of trades identified was an acceptable 281, which means</p><p>that you would not have been flooded with a barrage of stocks every</p><p>week. Some seem to think that a technique must produce a truckload of</p><p>signals each week in order to be effective. In reality, once you are fully</p><p>exposed in the market, just a few candidates each week are more than</p><p>enough to keep you fully committed.</p><p>Equity curve and yearly profits</p><p>Figure 8.18 shows the equity curve for the initial high strategy using a</p><p>weekly 34-period comparative strength channel. As can be seen in figure</p><p>8.18, the strategy performed in a relatively solid and very consistent manner</p><p>right from 2000. Although the equity curve did experience a temporary lull</p><p>from mid-2002 through to March 2003, this inactivity had little or no effect</p><p>of the Overall performance. A positive aspect of any pause in capital growth</p><p>is if there was a lack of any notable drawdown. During the lull period above,</p><p>the trading capital would have declined by $5102.86, which is a loss that is</p><p>well within manageable limits.</p><p>Figure 8.18: equity curve for initial high strategy using weekly 34-period</p><p>comparative strength channel</p><p>The equity curve indicates dependable capital growth Over an extended</p><p>period since the low of March 2003.</p><p>A look at the yearly profits, shown in figure 8.19 shows a picture of</p><p>consistency. The following annualised returns were produced:</p><p>Figure 8.19: yearly profits for initial high strategy using weekly 34-</p><p>period comparative strength channel</p><p>2000 — 15.5521 per cent</p><p>2001 — 43.7179 per cent</p><p>2002 — 50.2089 per cent</p><p>2003 — 18.7154 per cent</p><p>2004 — 19.6244 per cent</p><p>2005 — 30.432 per cent.</p><p>What can I say that the yearly profits chart does not already portray? The</p><p>degree of consistency in return had me repeating the testing process — even</p><p>though I knew deep down that the results were reliable, as I trade the</p><p>technique in real time. Interestingly,</p><p>the best returns came from trading on</p><p>strength during weaker conditions, such as experienced in 2001 and 2002. I</p><p>have suggested on many occasions that focusing on strength in weak or</p><p>bearish conditions is a more productive approach for those who wish to trade</p><p>from the long side; however, many feel that strength analysis is primarily</p><p>irrelevant. I have often heard the comment that it seems to make no</p><p>difference. This is not the case.</p><p>What you should be beginning to see from the results discussed in this</p><p>chapter is that channels used as confirmation tools and combined with a</p><p>suitable entry trigger can produce good results. However, the parameters for</p><p>the strategy need to be relative to trend development. Rather than just</p><p>accepting default settings, different values need to be tested so that the values</p><p>that achieve the best results can be determined. Changing the parameters so</p><p>that a 17-week initial high was combined with a weekly 34-period</p><p>comparative strength channel improved the results dramatically.</p><p>Coding for chapter 8</p><p>Comparative Strength Index — BullCharts</p><p>[Target=Percent; author=Wilson, Leon]</p><p>Symbol:=inputsymbol("Security","XJO");</p><p>Periods:=input("Periods",34,1,250);</p><p>Smoothing:=input("Smoothing", 1, 1, 55);</p><p>Trigger:=input("Trigger Line", 1, 1, 55);</p><p>Value2:=input("Over Bought", 70, 51, 99);</p><p>Value3:=input("Over Sold", 30, 1, 49);</p><p>Value4:=input("Upper Neutral Zone", 50, 50, 99);</p><p>Value5:=input("Lower Neutral Zone", 50, 1, 50);</p><p>Channel:=Close/LoadSymbol(symbol, Close);</p><p>Index:= Ma(RSI(Channel, Periods), Smoothing, E);</p><p>Trigg:=Ma(Ma(RSI(Channel, Periods), Smoothing, E),Trigger,E);</p><p>[Color=Black;Linestyle=Solid]</p><p>Index;</p><p>[Color=Blue;Linestyle=dash]</p><p>Trigg;</p><p>[Color=Light Slate Gray;Linestyle=Dotted]</p><p>value3;value2;</p><p>[name=1st Zone1]</p><p>Value5;</p><p>[name=2nd Zone2]</p><p>value4;</p><p>[name=Bullish Zone; linestyle=Fill;color=Light Slate Gray]</p><p>Value2; Value4;</p><p>[name=Neutral Zone; linestyle=Fill;color=White]</p><p>value5; Value4;</p><p>[name=Bullish Zone; linestyle=Fill;color=Dark Grey]</p><p>Value3; Value5;</p><p>Copyright Leon Wilson, 2005</p><p>Comparative Strength Channel — BullCharts</p><p>[Target=Price; author=Wilson, Leon]</p><p>Ticker:=inputsymbol("Security","XJO");</p><p>Periods:=input("Channel Periods", 34, 1, 250);</p><p>Smoothing:=input("Smoothing", 1, 1, 55);</p><p>Value2:=input("Over Bought", 70, 50, 99);</p><p>Value3:=input("Over Sold", 30, 1, 50);</p><p>Value4:=input("Upper Neutral Zone", 55, 50, 99);</p><p>Value5:=input("Lower Neutral Zone", 45, 1, 50);</p><p>Channel:=Close/LoadSymbol(Ticker, Close);</p><p>[Color=Light Slate Gray;Linestyle=Dotted]</p><p>OB:=Ma((RSI(Channel,Periods)-Value3),Smoothing,E);</p><p>OS:=Ma((RSI(Channel,Periods)-Value2),Smoothing,E);</p><p>NZU:=Ma((RSI(Channel,Periods)-Value4),Smoothing,E);</p><p>NZL:=Ma((RSI(Channel,Periods)-Value5),Smoothing,E);</p><p>{Show Results}</p><p>[name=OverB;]</p><p>Close-(Close*(OB/100));</p><p>[name=OverS;]</p><p>Close-(Close*(OS/100));</p><p>[name=NeutUp;]</p><p>Close-(Close*(NZU/100));</p><p>[name=NeutLower;]</p><p>Close-(Close*(NZL/100));</p><p>{Upper Channel}</p><p>[name=Upper; linestyle=Fill;color=Light Slate Gray]</p><p>Close-(Close*(NZU/100)); Close-(Close*(OS/100));</p><p>{Neutral Zone}</p><p>[name=Zone; linestyle=Fill;color=White]</p><p>Close-(Close*(NZU/100)); Close-(Close*(NZL/100));</p><p>{Lower Channel}</p><p>[name=Lower ; linestyle=Fill;color=Dark Grey]</p><p>Close-(Close*(NZL/100));Close-(Close*(OB/100));</p><p>Copyright Leon Wilson, 2005</p><p>Comparative Strength Index — MetaStock</p><p>Periods:=Input("Periods", 1,250,34);</p><p>Smoothing:=Input("Smoothing", 1, 55, 1);</p><p>Value2:=Input("Over Bought", 50, 99, 70);</p><p>Value3:=Input("Over Sold", 1, 50, 30);</p><p>Value4:=Input("Upper Neutral Zone", 50, 99, 55);</p><p>Value5:=Input("Lower Neutral Zone", 1, 50, 45);</p><p>BM:=Security("C:\My Databases\MetaStock\ASX\X0\XJO",CLOSE);</p><p>Ratio:=CLOSE/BM;</p><p>Index:= Mov(RSI(Ratio,Periods),Smoothing,E);</p><p>Trigg:=Mov(Mov(RSI(Ratio,Periods),Smoothing,E),1,E);</p><p>Index; Trigg;</p><p>Value3; Value2; Value5; Value4;</p><p>Copyright Leon Wilson, 2005</p><p>Comparative Strength Channel — MetaStock</p><p>Periods:=Input("Channel Periods", 1, 250, 34);</p><p>Smoothing:=Input("Smoothing", 1, 55, 1);</p><p>Value2:=Input("Over Bought", 50, 99, 70);</p><p>Value3:=Input("Over Sold", 1, 50, 30);</p><p>Value4:=Input("Upper Neutral Zone", 50, 99, 55);</p><p>Value5:=Input("Lower Neutral Zone", 1, 50, 45);</p><p>BM:=Security("C:\My Databases\MetaStock\ASX\X0\XJO",CLOSE);</p><p>Ratio:=CLOSE/BM;</p><p>Channel:= RSI(Ratio,Periods);</p><p>OB:=Mov((RSI(Channel,Periods)-Value3),Smoothing,E);</p><p>OS:=Mov((RSI(Channel,Periods)-Value2),Smoothing,E);</p><p>NZU:=Mov((RSI(Channel,Periods)-Value4),Smoothing,E);</p><p>NZL:=Mov((RSI(Channel,Periods)-Value5),Smoothing,E);</p><p>CLOSE-(CLOSE*(OB/100));</p><p>CLOSE-(CLOSE*(OS/100));</p><p>CLOSE-(CLOSE*(NZU/100));</p><p>CLOSE-(CLOSE*(NZL/100));</p><p>Copyright Leon Wilson, 2005</p><p>Comparative Strength Channel — TradeSim (MetaStock)</p><p>Pds:=34;</p><p>Comp:=CLOSE/INDICATOR;</p><p>Index:=RSI(Comp,Pds);</p><p>NZU:=Index -55;</p><p>NZL:=Index -45;</p><p>UC:=Index -70;</p><p>NeutUp:=CLOSE-(CLOSE*(NZU/100));</p><p>OB:=CLOSE-(CLOSE*(UC/100));</p><p>NeutL:= CLOSE-(CLOSE*(NZL/100));</p><p>EntryTrigger := CLOSE>NeutUp AND Ref(CLOSE,-1)<=Ref(NeutUp,-1);</p><p>EntryPrice := OPEN;</p><p>ExitTrigger := (CLOSE<NeutL AND (Ref(CLOSE,-1) >=Ref(NeutL,-1))) OR</p><p>(CLOSE<OB AND (Ref(CLOSE,-1) >=Ref(OB,-1)));</p><p>ExitPrice := OPEN;</p><p>InitialStop:=0;</p><p>Copyright Leon Wilson, 2005</p><p>Initial high Indicator — BullCharts</p><p>[Target=Price; author=Wilson, Leon]</p><p>IH:=input("Initial High", 21, 1, 300);</p><p>HH:=IH-1;</p><p>If(Ref(HIGH,-HH)>=HHV(HIGH,HH),HHV(HIGH,HH),If(Ref(HIGH,-HH)<HHV(HIGH,H</p><p>H),PREV,0));</p><p>Copyright Leon Wilson, 2005</p><p>Initial high Indicator — MetaStock</p><p>IH:=Input("Initial High",1,300,21);</p><p>HH:=IH-1;</p><p>If(Ref(HIGH,-HH)>=HHV(HIGH,HH),HHV(HIGH,HH),If(Ref(HIGH,-HH)<HHV(HIGH,H</p><p>H),PREV,0));</p><p>Copyright Leon Wilson, 2005</p><p>Initial high/Directional Price Channel on Rel% Stop — TradeSim</p><p>HH:=21;</p><p>SetUp:=If(Ref(HIGH,-HH)>=HHV(HIGH,HH),HHV(HIGH,HH),If(Ref(HIGH,-HH)<HHV(</p><p>HIGH,HH),PREV,0));</p><p>Pds:=55;</p><p>Index:=100-(100/(1+(PDI(Pds)/(MDI(Pds)+0.0001))));</p><p>NZU:=Index -55;</p><p>NeutUp:=CLOSE-(CLOSE*(NZU/100));</p><p>ValueA:=(CLOSE-(CLOSE-(((Sum(CLOSE-LOW,21)/21)+ATR(21)))))/CLOSE;;</p><p>Stop:=CLOSE-((CLOSE*ValueA)*2.7);</p><p>Entry:=Cross(CLOSE,SetUp) AND CLOSE>NeutUp; Trigg:=HighestSince(1,Entry,</p><p>Stop);</p><p>EntryTrigger :=Cross(CLOSE,SetUp) AND CLOSE>NeutUp;</p><p>EntryPrice := OPEN;</p><p>ExitTrigger:=CLOSE<Trigg AND (Ref(CLOSE,-1)>=Ref(Trigg,-1));</p><p>ExitPrice := OPEN;</p><p>InitialStop:=0;</p><p>Copyright Leon Wilson, 2005</p><p>Chapter 9</p><p>Bilateral relationships</p><p>The discussions in the previous chapters have been fairly general, and have</p><p>primarily focused on the application of relative parameters. Testing revolved</p><p>around the specific behaviour of the closing price in relation to the bullish</p><p>and bearish bands. The adaptation of the relative strength index and similar</p><p>indicators to price action can also expose you to a range of occurrences with</p><p>which you may be unfamiliar. Much of what is discussed in this chapter</p><p>builds on the basic foundation set in place through the previous chapters.</p><p>Crossing the neutral zone — bullish</p><p>Price action behaves in an unusual manner when it attempts to transcend the</p><p>neutral zone. It is relatively common for trending price action to cross the</p><p>neutral zone, perhaps leading traders to believe that the worst is behind them,</p><p>only to stop and retrace. I have always had a policy of trading weak stocks</p><p>from the short side and strong stocks from the long side; however, since I</p><p>began using relative price channels with my analysis, I have become</p><p>increasingly hesitant about riding with the transition from bullish to bearish</p><p>and vice versa. The neutral zone deserves spooky music at times. If ever there</p><p>was a Bermuda triangle in price action, this little region would have to be it.</p><p>It’s almost as though a stock is not permitted to switch from bearish to bullish</p><p>characteristics without first pausing. The stock seems to stop in the neutral</p><p>zone while all of the long traders climb on board and all the short traders</p><p>disembark. A clear and decisive transition directly from strong bearish to</p><p>strong bullish conditions or vice versa is not that common.</p><p>Tip</p><p>Never become so focused on indicator behaviour that you lose sight of price action.</p><p>Combining price action with channel</p><p>development reveals many consistencies within the</p><p>market. Understanding the market and how it responds also allows you to understand</p><p>probability. Understanding probability comes from understanding the many bilateral</p><p>relationships that exist within the market.</p><p>Figure 9.1 highlights the typical behaviour of price action when it attempts to</p><p>move from the bearish band into the bullish region of the price channel.</p><p>Figure 9.1: price action as it moves from bearish to bullish conditions</p><p>The price action of Fairfax clearly highlights this behaviour, as can be seen in</p><p>figure 9.2.</p><p>Figure 9.2: Fairfax price action and neutral zone crossOvers</p><p>In the first boxed area in figure 9.2, price action gravitated toward the top half</p><p>of the bearish band before commencing a brief bullish rally. The rally</p><p>managed to enter the lower portion of the bullish band only to pause. It failed</p><p>to produce any meaningful bullish values before consolidating, and finally</p><p>retraced.</p><p>The second boxed area in figure 9.2 shows a solid bearish rally that tested</p><p>the upper cord of the bearish band before rebounding. It failed to produce any</p><p>meaningful bearish values before returning to the top half of the neutral zone.</p><p>The price action of Fairfax is consistent with the price action of a large</p><p>percentage of stocks. When price action rises from bearish to bullish, seldom</p><p>will the stock go on to tag the upper cord of the bullish band without first</p><p>pausing in the process. Price action will first tag the lower cord of the bullish</p><p>band and then pause or return to the centre of the neutral zone before</p><p>continuing with up-trending activity.</p><p>Based on this, the following guidelines can be applied for interpretation:</p><p>For price action that is rising from the bearish band, look for price action</p><p>to pause or retrace once it tags the lower cord of the bullish band.</p><p>When a primary trend swings from bearish to bullish, look for price</p><p>action to initially rally, pause or retrace slightly, then trend. (This is</p><p>consistent with the accepted notion of trend behaviour.)</p><p>In the circled area on figure 9.2, I have also highlighted another regular</p><p>occurrence. This builds on the principle that a stock will seldom continue to</p><p>trend having crossed the neutral zone without first pausing when it meets the</p><p>opposing band. Bullish rallies usually commence when price action is either</p><p>in the centre or top half of the neutral zone. Being able to define whether</p><p>consolidation periods are predominantly bullish or bearish also allows you to</p><p>determine the probable direction of price action prior to the eventual break-</p><p>out. This leads to the next guidelines:</p><p>Successful bullish rallies will usually commence when price action</p><p>develops in either the centre or the top half of the neutral zone.</p><p>Successful bullish rallies that commence from the bottom half of the</p><p>neutral zone have usually experienced a period of notable consolidation</p><p>in the lead-up to trending activity.</p><p>In order to clarify the discussion so far, I will look more closely at price</p><p>action, using the chart of ANZ shown in figure 9.3. This time transition</p><p>behaviour is combined with entry signals.</p><p>Figure 9.3: ANZ price action and band rebounds and break-outs</p><p>In figure 9.3, I have highlighted the following areas of interest:</p><p>Box 1 — another stock, the same outcome. Price action declined from</p><p>the bullish band to successfully cross the neutral zone, only to rebound</p><p>in an upward manner following a tag of the upper cord of the bearish</p><p>band. Interestingly, the rebound tagged the lower cord of the upper band</p><p>to again pause. It is almost as though there were opposing forces at work</p><p>that must be neutralised before trending activity can resume. Note that a</p><p>solid penetration of the bearish band was following by a minimal</p><p>penetration of the bullish band.</p><p>Box 2 — although this box looks similar to the first box, there is a subtle</p><p>difference. Price action continued to bounce around within the confines</p><p>of the neutral zone, but the bullish rally was much stronger this time</p><p>around. The bearish activity that immediately preceded the rally is</p><p>greatly reduced, which means the potential for greater upside exists.</p><p>Note that a minimal penetration of the bearish band was followed by the</p><p>solid penetration of the bullish band.</p><p>Box 3 — this box shows what later proves to be a solid rally and the</p><p>early signs of trend development. The rally commenced well within the</p><p>confines of the neutral zone. The difference here is that price action</p><p>never threatened to penetrate the bearish band at any stage. After price</p><p>action dropped from the bullish band into the neutral zone, it gravitated</p><p>toward the top half of the neutral zone before breaking to the upside.</p><p>Are you starting to see how price action relates to the price bands?</p><p>Box 4 — in this box, price action drops briefly into the neutral zone,</p><p>only to rebound. The trend continues unabated at the time of writing.</p><p>There is little or no evidence to suggest the presence of any bearish</p><p>influence.</p><p>Before I examine each entry signal, remember that I identified the first</p><p>closing price that occurred above the lower cord of the bullish band as the</p><p>entry trigger. This is the same trigger that has been applied throughout the</p><p>benchmarking process.</p><p>The following entry signals have been highlighted:</p><p>Entry signal 1 — using the principles of price action never crossing the</p><p>divide without first pausing, the first entry signal is potentially the</p><p>weakest development in price action. At this point, price action initially</p><p>dropped into the bearish zone of trend development, rebounded to</p><p>successfully penetrate the lower cord of the bullish band but then stalled.</p><p>The more I study this style of relative analysis, the more times I see rally</p><p>failure when price action crosses the neutral divide. This is not an ideal</p><p>entry signal and one that you should look to avoid due to the notable</p><p>penetration into the bearish band immediately prior.</p><p>Entry signal 2 — for the second entry signal, price action closes within</p><p>the bullish band. While price action has again rebounded from the</p><p>bearish band, previous penetration into the bearish region this time was</p><p>minimal. Price action quickly fails and promptly returns to the neutral</p><p>zone. Again this is not a strong entry signal as price action has</p><p>rebounded from the bearish band without pausing in the process. While</p><p>the initial penetration of the bearish band was not as significant, the</p><p>eventual outcome was similar. Although price action did manage to push</p><p>into the bullish band a little further than previously, you should also</p><p>avoid this type of entry signal.</p><p>Entry signal 3 — price action breaks above the lower cord of the bullish</p><p>band, but this time rallies nicely to eventually tag the upper cord of the</p><p>bullish band before retracing. Interesting enough, price action dips to</p><p>almost tag the lower cord of the bullish band. When it comes to a</p><p>legitimate entry signal, this is the type of set-up you should look for.</p><p>Although the trigger was generated following brief activity in the neutral</p><p>zone, price action was not in the process of rebounding from the bearish</p><p>band.</p><p>Entry signal 4 — price action dips into the neutral zone briefly, only to</p><p>rebound in a strong bullish manner. The break-out was on good range</p><p>with the closing price approaching the high. Again price action</p><p>continues on to tag the outer cord of the bullish band.</p><p>Can you begin to see how the strength of confirmation signals relates to</p><p>previous behaviour? When traders apply the RSI in its original format at the</p><p>bottom of the screen, it is impossible to know how previous activity is</p><p>relative to strength and the legitimacy of the confirmation signal as the</p><p>indicator swings above 55 per cent.</p><p>Long-term bullish behaviour</p><p>So far I have examined basic entry conditions as price action enters the</p><p>bullish band and how you can determine whether an entry signal is</p><p>potentially legitimate by analysing lead-up activity prior to an initial upward</p><p>move. However, what about a trend that is well underway? A common belief,</p><p>based on conventional analysis,</p><p>suggests that once the RSI drops below 50</p><p>per cent, it represents a bearish environment. This is an incorrect assumption,</p><p>as it assumes that the divide that separates bullish and bearish activity can be</p><p>identified as a definitive point in trend development, which is not the case.</p><p>I’m not big fan of showing perfect examples, as they are seldom</p><p>encountered in real time; however, Bendigo Bank highlights general trend</p><p>development so clearly, it would be impossible to achieve the same clarity</p><p>with the kind of basic drawing shown in figure 9.4, or with pages of written</p><p>text.</p><p>Figure 9.4: price action during a rising trend</p><p>While it is essential to have a mental picture of basic behaviour, you should</p><p>not ignore a pristine example when it occurs. Take the time to study the chart</p><p>shown in figure 9.5 and to examine where price action commences for each</p><p>rally in relation to the relative price channel, and how it responds once the</p><p>rally concludes. Also take the time to study price action in the lead-up to the</p><p>rally commencing — for example, was it consolidating or on the rebound?</p><p>They say a picture is worth a thousand words — Bendigo Bank is worth</p><p>considerably more if you take the time to study it closely.</p><p>Figure 9.5: trend continuation</p><p>Bendigo Bank displays the classic signs of sustainable trend development. I</p><p>have highlighted where rally behaviour commenced and then, for all practical</p><p>purposes, tagged the upper cord of the bullish band. Price action then</p><p>retreated back to neutral territory. For each occasion where price action</p><p>tagged or almost tagged the upper cord of the bullish band, have a look where</p><p>each rally commenced (the boxed areas on figure 9.5). As can be seen, each</p><p>rally commenced from approximately the centre of the neutral zone or above.</p><p>The following points look at each of the areas highlighted on figure 9.5 in</p><p>more detail:</p><p>Box 1 — in the first boxed area, price action consolidated within the</p><p>neutral zone and at some distance from the upper cord of the bearish</p><p>band. While it may not have looked this obvious in real time — as the</p><p>rising trend has compressed earlier price action, thereby enhancing the</p><p>impression that this was consolidating price action — the same</p><p>principles apply. Price action broke upward after developing in the top</p><p>two-thirds of the neutral zone. The rally briefly paused in the lower half</p><p>of the bullish band, confirming the presence of a bullish consolidation.</p><p>Price action finally bounced off the lower cord of the bullish band to</p><p>break to the upside, almost tagging the upper cord of the bullish band.</p><p>As price action failed to tag the upper cord of the bullish band, the</p><p>ensuing retracement was not severe and concluded around the middle of</p><p>the neutral zone. (Try to grasp the concept of equal and opposite forces</p><p>when it comes to supply and demand. Often one extreme action is</p><p>followed by equally extreme action.)</p><p>Box 2 — the second rally again commenced in the middle of the neutral</p><p>zone, and price action rallied strongly to finally tag the upper cord of the</p><p>bullish band. Price action immediately retraced to test the lower cord of</p><p>the bullish band, where it was met by bullish support and the decline</p><p>halted. Price action drifted into the neutral zone where the low</p><p>eventually tagged the upper cord of the bearish band. While the bearish</p><p>penetration was not severe, its presence is notable.</p><p>Circled — the next rally commenced from a rebound immediately</p><p>following the penetration of the bearish band. The rally comfortably</p><p>failed to tag the upper cord of the bullish band. While the distance from</p><p>the upper cord of the bullish band at which the rally concluded does not</p><p>appear significant on the chart shown in figure 9.5, when forward data is</p><p>removed it can be seen that the rally comfortably failed to threaten the</p><p>upper cord. This rally concluded on a high of $8.52, while the upper</p><p>cord had a value of $8.92 at that time. The following retracement</p><p>concluded with a tag of the upper cord of the bearish band. Even with</p><p>bullish stocks that are trending Over the longer term, a stock that has</p><p>moved from the bearish channel to the bullish channel will seldom run</p><p>onto tag the upper cord of the bullish band.</p><p>Box 3 — price action develops in a slightly different manner with the</p><p>rally labelled in the third box. Following the rebound off the upper cord</p><p>of the bearish band, price action then consolidated in the middle of the</p><p>neutral zone. The initial consolidation period was immediately followed</p><p>by a second consolidation period that developed in the top half of the</p><p>neutral zone. This behaviour nullifies the influence any previous</p><p>rebound may have had. Price broke to the upside and proceeded to test</p><p>the upper cord of the bullish band, but it fell just shy of making solid</p><p>contact. The retracement activity that followed concluded in the top half</p><p>of the neutral zone.</p><p>Box 4 — are you starting to see repetitive behaviour unfolding before</p><p>you? The rally highlighted in the fourth box commenced from a</p><p>consolidation period that developed within the top half of the neutral</p><p>zone. Price action rallied hard to quickly reach the upper cord of the</p><p>bullish band. This is no longer a surprise is it? I’ll bet you never worked</p><p>out this type of information from your conventional index on the bottom</p><p>of your chart. Price action retreated into the lower cord of the bullish</p><p>band before eventually declining into the neutral zone. The retracement</p><p>concluded with price action consolidating on the upper cord of the</p><p>bearish band.</p><p>Box 5 — the rally shown at box five (still continuing at the time of</p><p>writing) is naturally sustainable. While I have to be careful about</p><p>making predictions, there was a strong period of consolidating price in</p><p>the middle of the neutral zone prior to renewed trending activity. An</p><p>immediate rebound following the tag of the upper cord of the bearish</p><p>band wouldn’t have been a good sign. Price action not only consolidated</p><p>upon the conclusion of retracement activity, it again paused relatively</p><p>centrally in the neutral zone prior to finally breaking upward. What do</p><p>you think the chances are of price action eventually testing the upper</p><p>cord of the bullish band? Ultimately, only time will tell.</p><p>If you compare the starting point of each rally and the conclusion of the</p><p>ensuing retracement, each usually developed from or returned to in a similar</p><p>position in the neutral zone. While it does not occur with every retracement,</p><p>it happens all too frequently. This is pretty handy information for the short</p><p>trader, don’t you think?</p><p>Four of the five break-outs shown in figure 9.5 that went onto effectively</p><p>tag the upper cord of the bullish band also commenced from a minor price</p><p>cluster that developed either in the middle of or within the top half of the</p><p>neutral zone. By evaluating how price action relates to the relative price</p><p>channel, you can begin to develop an understanding of how to identify</p><p>potential signals of significance. Much of the indicator behaviour that</p><p>develops between prominent highs and lows is not indicator noise but a</p><p>reflection of market forces at work. For the first time, more than just</p><p>generalised indicator behaviour can be understood.</p><p>So what have we discOvered so far? The following points can be made:</p><p>Sustainable rallies seldom start following a rebound from the bearish</p><p>band.</p><p>Longer term trends will regularly pause in the neutral zone before</p><p>continuing. The accepted notion that you should never hold a stock from</p><p>the long side when the RSI is less than 50 per cent is therefore incorrect.</p><p>The conclusion of retracement activity is often in close proximity to the</p><p>starting point of the preceding rally in relation to its positioning within</p><p>the neutral zone — that is, if the rally commences from the middle of the</p><p>neutral zone, the ensuing retracement will generally conclude in the</p><p>middle of the neutral zone. This is primarily applicable to those stocks in</p><p>established trends — although similar behaviour can be seen across a</p><p>range of conditions.</p><p>Crossing the neutral zone — bearish</p><p>So far the focus</p><p>has been on bullish activity; however, a similar relationship</p><p>exists between trend development and the relative price channel when a stock</p><p>becomes bearish. As a stock moves downwards, price action will decline</p><p>from the bullish band until it tags the upper cord of the bearish band. This is</p><p>then usually followed by a rebound to the middle to top half of the neutral</p><p>zone, after which price action will usually pause before continuing with</p><p>down-trending activity. This is shown in figure 9.6. As I have already</p><p>cOvered this in some detail for bullish activity, I will discuss bearish activity</p><p>more briefly. Note that if price action is declining from the bullish band, look</p><p>for price action to pause or retrace once it tags the upper cord of the bearish</p><p>band.</p><p>Figure 9.6: declining price action</p><p>At the time of writing, Gunns Limited (GNS) had recently experienced a drop</p><p>in price action following strong bullish activity Over a sustained period. The</p><p>same behavioural patterns as discussed above, but in reverse, can be</p><p>identified. While it may appear that I am simply repeating the same</p><p>information — which is partially true, as I want the concept of outer cord</p><p>penetration followed by retracement to the opposing cord of the neutral zone</p><p>to become cemented in your psyche — I also want to highlight how this</p><p>occurrence tends to repeat itself regardless of previous trend development.</p><p>On figure 9.7, I have highlighted the final two retracements leading to the</p><p>eventual reversal of GNS. In order to more fully appreciate the final decline, I</p><p>will look at both retracements as the price action that led to their development</p><p>differed significantly.</p><p>Figure 9.7: retracement following a rally</p><p>The following areas have been highlighted on figure 9.7:</p><p>A Trend development leading into the initial retracement unfolded in a</p><p>consistent and stable manner, driven by constant momentum and low</p><p>volatility. The eventual bullish penetration of the upper cord of the bullish</p><p>band was minimal. As expected, although the ensuing retracement briefly</p><p>tagged the upper cord of the bearish band, the band was never under any</p><p>real threat of severe penetration. Price action then began to cluster in the</p><p>top half of the neutral zone.</p><p>B The second rally developed in a manner contrary to the first, with</p><p>momentum and volatility exponentially increasing as the trend reached a</p><p>point of unsustainability. The bullish penetration of the upper cord of the</p><p>bullish band was severe and sustained. I have refrained from using the</p><p>phrase ‘trend maturity’, as I did not know this at the time; however, there</p><p>is certainly sufficient evidence in the chart to suggest that a reversal of</p><p>some degree was imminent. It quickly became apparent that no</p><p>consolidation period was to follow, suggesting that the uptrend was</p><p>probably Over. Severe bullish price action was instead mirrored by a</p><p>strong decline, with price action comfortably penetrating the upper cord</p><p>of the bearish band before pausing.</p><p>By the time price action tagged the upper cord of the bearish band, it was</p><p>obvious that this retracement behaviour differed significantly from the</p><p>previous development. The sell-off was strong and unrelenting and, unlike</p><p>during bullish retracement behaviour, price action had little trouble in</p><p>reaching the upper cord of the bearish band. If you flick back to the price</p><p>action of Bendigo Bank shown in figure 9.5, which highlighted retracement</p><p>behaviour in up-trending conditions, you will notice that each time price</p><p>action approached the upper cord of the bearish band, it did not penetrate it in</p><p>a convincing manner. In that case, price action retraced and then paused in</p><p>the middle of the neutral zone. The final dip in price was brief following a</p><p>pause in activity, and any bearish activity was generally weak. Not once did</p><p>the price action of Bendigo Bank drop from the upper region of the bullish</p><p>band to immediately tag the upper cord of the bearish band.</p><p>Now have a look at the chart of Gunns. Following the final rally, price</p><p>action declined without experiencing a pause in the process to easily</p><p>penetrate the bearish band.</p><p>With price action dropping into the bearish band, would you consider</p><p>taking a long position? Of course not. However, conventional indicator</p><p>values often lead to the belief that trading a rebound such as this is a sound</p><p>approach. The channel puts price action and general trend development into</p><p>perspective.</p><p>As can be seen in figure 9.8, the wheels spectacularly detached from the</p><p>Gunns’ cart, driven by a solid bearish rally. While it caught some by surprise,</p><p>the channel provided plenty of warning that a potential reversal was in the</p><p>wind.</p><p>Figure 9.8: bearish decline</p><p>The following points can be highlighted from figure 9.8:</p><p>Price action rebounded off the upper cord of the bearish band and, guess</p><p>what? It tagged the opposing band of the neutral zone. The rebound</p><p>failed as expected, but price action also failed to hold the middle region</p><p>of the neutral zone.</p><p>Price action then entered the bearish band where it declined to tag the</p><p>lower cord of the bearish band and then, guess what? It rebounded. This</p><p>is all becoming a little repetitious isn’t it? The degree of consistency in</p><p>trend development when analysed in conjunction with relative analysis</p><p>is similar to Newton’s law of relativity. Every force has an equal and</p><p>opposite force. The same behaviour repeats itself with uncanny accuracy</p><p>Over a wide-ranging number of stocks.</p><p>Long-term bearish behaviour</p><p>Not all stocks developing in a down-trending manner experience price action</p><p>that occurs within the lower half of the bearish band as you might normally</p><p>expect. This is one common difference that I have identified between up-</p><p>trending and down-trending stocks. Seldom does a bullish trend develop</p><p>while price action remains within the confines of the neutral zone. Sure, brief</p><p>rallies will occur, but for an uptrend to have become sustainable, price action</p><p>will usually have had to move into the bullish band, and often toward the</p><p>upper cord.</p><p>For a percentage of down-trending stocks, price action will barely move</p><p>below the upper cord of the bearish band. When I initially noticed this</p><p>occurrence, I was somewhat perplexed as to why sustainable uptrends</p><p>required price action to move out of the neutral band, and often well into the</p><p>bullish band, but sustainable downtrends could occur while experiencing only</p><p>minimal bearish activity. The only conclusion I could draw was that no action</p><p>is required from the buyers in order to drive price action downward.</p><p>When it comes to an up-trending stock, traders who do not own the stock</p><p>must initiate the transaction process by going to the market and placing an</p><p>order. The actions are twofold here — the trader must bid for the stock and</p><p>the seller must accept. Only the most nervous, uneducated and inexperienced</p><p>(or desperate) sell into a rising market. The buyers must become aggressive if</p><p>they going to persuade most owners to part with their holdings. As buyers are</p><p>unable to hire goons to pay the current owners a visit, they have to offer a</p><p>higher price instead (which is probably a good thing).</p><p>When it comes to a down-trending stock, no action is required from the</p><p>buyers. The sellers go to the market and take out the bids sitting in the</p><p>system. The key difference here is that the buyers are not actively chasing the</p><p>stock. They may include the ever-optimistic bottom fishers, who sternly</p><p>believe they can pick the bottom of the market, and long-term investors, who</p><p>are just happy to buy at a slightly cheaper price in the belief that they have</p><p>just picked up a bargain. While the steady flow of sellers consistently takes</p><p>out the bargain hunters, there is little enthusiasm from the serious buyers.</p><p>Up-trending and down-trending behaviour can be summarised as follows:</p><p>For uptrends, there are eager, educated buyers and reluctant sellers.</p><p>For downtrends, there are uninterested educated buyers, concerned</p><p>sellers and uneducated bargain hunters.</p><p>Figure 9.9 shows typical price action during a downtrend.</p><p>Figure</p><p>9.9: declining trend</p><p>While bargain hunters will seldom chase an upward trend, they will happy</p><p>follow a trend downward, throwing good money after bad. This is the</p><p>opposite of the actions of the educated trader. It is the continued presence and</p><p>steady stream of bargain hunters and long-term investors that allows</p><p>downward trends to steadily develop.</p><p>In a bullish trend, those who have the stock, want it, and those who don’t</p><p>have it, want it also. In a bearish trend, those who have the stock, don’t want</p><p>it, and those who don’t have it, don’t want it either.</p><p>In reality, it is largely immaterial as to why sustained down-trending activity</p><p>can develop on minimal bearish activity. All you need to know is that it</p><p>happens — and often with a fair degree of regularity. In order to explain this</p><p>in more detail, I will look at the price action of Fairfax, which is one of many</p><p>stocks that have declined on minimal bearish behaviour.</p><p>As shown in figure 9.10, although Fairfax has declined since August 2000,</p><p>price action predominately remained within the confines of the neutral zone</p><p>or the upper region of the bearish band. The Oversold region was never really</p><p>threatened by declining price action between 2000 and 2003, clearly</p><p>indicating that strong bearish activity is not essential for declining price</p><p>action to eventuate.</p><p>Figure 9.10: down-trending on neutral bias</p><p>Conventional theory suggests that declining stocks will produce low RSI</p><p>values. This is stating the obvious, as this must be the case under certain</p><p>conditions because of the calculation process; however, this is also an</p><p>assumption that leads people to believe that all bearish RSI behaviour will</p><p>mirror bullish activity. The simple fact is a strong bearish presence is not</p><p>required in order for a stock to decline. This is nowhere more clearly evident</p><p>than with the price action of Fairfax.</p><p>The following areas have been highlighted on figure 9.10:</p><p>Box 1 — in this area, there was the usual build-up of price action in the</p><p>lower range of the neutral zone. There is a minor difference here when</p><p>this is compared with bullish activity. Bullish activity that consolidates</p><p>tends to remain relatively static; however price action during the</p><p>consolidation period here is walking down the upper cord of the bearish</p><p>band. This is a relatively common occurrence, so in this situation, you</p><p>need to check the underlying direction of the price channel. In this case,</p><p>the channel was declining, suggesting that the neutral price action had a</p><p>bearish influence. A distinguishable bearish feature is the range</p><p>associated with the price action itself in this area. If you look to the left</p><p>of the chart, you will notice a similar piece of price action. It is also</p><p>consolidating as it walks along the upper cord of the bearish band. The</p><p>difference here is that the trading range is reduced while there is no</p><p>obvious decline in channel behaviour. While you must be careful when</p><p>comparing price to the left of the screen, understanding the often subtle</p><p>differences in regard to ongoing development will allow you to separate</p><p>bullish from bearish price action, even when it appears similar on the</p><p>surface. In box one, price action finally broke to the downside, where it</p><p>threatened, but failed, to tag the lower cord of the bearish band.</p><p>Box 2 — having failed to tag the lower cord of the bearish band, price</p><p>action rebounded; however, it failed to significantly tag the lower cord</p><p>of the bullish band upon its return. Price action remained in the bottom</p><p>half of the neutral zone for the majority of retracement activity.</p><p>Knowing what you have learned so far, would you take a long position</p><p>on the first signs of bullish retracement activity? Of course not. Bottom</p><p>fishing doesn’t seem so appealing when you analyse price action from</p><p>this perspective, does it? Especially now that you have the ability to</p><p>gauge a point of probable rally maturity once a position has been</p><p>opened. Those traders who go bottom fishing believe that they can hook</p><p>into a reversal stock that will travel to the heavens, eliminating the</p><p>impact of all previous losses. However, any bullish activity in a</p><p>downtrend will in all probability be short-lived, as it must transcend the</p><p>neutral zone.</p><p>Box 3 — a solid downward rally developed as price action declined</p><p>from the $4.00 region to around $2.60. This is a healthy and consistent</p><p>decline that any short trader would be happy with, yet price action</p><p>barely made it into the bearish band for the bulk of the decline. The final</p><p>drop in value had price action threatening the lower cord of the bearish</p><p>band; however, it was unable to penetrate the lower cord. As expected,</p><p>the ensuing retracement returned to the neutral zone. Another blow has</p><p>just been dealt to those eager to dredge the bottom for trading</p><p>opportunities.</p><p>Tailoring indicators to suit bear rallies</p><p>This leads me to another conclusion that directly affects shorting strategies.</p><p>When constructing a long side strategy, the majority of traders will include</p><p>bullish indicator parameters as a means of confirmation. When it comes to</p><p>short side strategies, the same principles are applied without question. What</p><p>is not realised by the majority of traders who are exploring the world of</p><p>shorting is that, as discussed above, down-trending activity does not require</p><p>strong bearish behaviour in order to occur or be sustained Over a long period.</p><p>The shorting strategy fails to produce because unrealistic confirmation</p><p>parameters were set.</p><p>Figure 9.11 again shows the declining price action of Fairfax.</p><p>Figure 9.11: repetitive behaviour</p><p>In figure 9.11, I have repeated the previous chart to avoid making the chart</p><p>excessively cluttered. (My apologies to my long-suffering editor.)</p><p>The following areas can be highlighted:</p><p>1 If you study price action to the left of the chart, you can see a period of</p><p>steadily increasing congestion that developed in the lower half of the</p><p>neutral zone. This progressively formed into a wedge formation before</p><p>price action eventually broke to the upside. If this behaviour is compared</p><p>to the activities of price action to the right, a remarkable similarity can be</p><p>seen.</p><p>2 Congesting price action, as highlighted at the points labelled with ‘1’,</p><p>on its own may for some be insufficient information to suggest that a</p><p>change in direction is in the wind. However, now examine the actions of</p><p>the price channel on the left. The price channel declined then plateaued,</p><p>forming a bottom. Again, the same characteristics developed on the far</p><p>right of the chart. The price channel declined, but price action did not take</p><p>out the previous low by declining further before pausing. This in turn</p><p>caused the price channel to plateau.</p><p>3 I have taken the liberty of highlighting how the price channel behaves</p><p>when reversals occur in trending activity in order to help build your</p><p>understanding. As price action reversed, this was mirrored by the price</p><p>channel. There was no lag between price action and the channel, with</p><p>both turning in unison.</p><p>Based on the price action shown in the far right of the chart, it would be</p><p>logical to expect price action to next break to the upside. Now another</p><p>question for you. Do you think you could derive the same information from</p><p>the conventional index at the bottom of the chart? Never forget to look</p><p>beyond the relationship that exists between price and the relative channel.</p><p>Often the characteristics of the price channel tell you as much about trend</p><p>development as the position of the closing price.</p><p>This leads me to the next point of interest — joint behaviour.</p><p>Understanding complex behaviour</p><p>The relationship that evolves between price action and price channels can</p><p>become quite complex once you begin to understand what is before you.</p><p>Often the more widely known approaches involve analysing the Overall</p><p>relationship that exists between an indicator and price action. The most</p><p>common occurrence that the majority of traders watch for is, without doubt,</p><p>divergences. While divergences can be a strong indication of pending price</p><p>action, very few</p><p>than yesterday</p><p>is also above the 125-day moving average.</p><p>The MetaStock coding for this would appear as follows:</p><p>Range:=((CLOSE-Ref(CLOSE,-1))/Ref(CLOSE,-1))*100;</p><p>(CLOSE>Mov(CLOSE,21,E)) AND (Ref(CLOSE,-1)<=Ref(Mov(CLOSE,21,E),-1))</p><p>AND CLOSE>Mov(CLOSE,125,E) AND Range>=5;</p><p>An exit trigger occurs when the close occurs below the 21-day moving</p><p>average for the first time.</p><p>The MetaStock coding for this would appear as follows:</p><p>(CLOSE<Mov(Close,21,E)) AND (Ref(Close,-1)>=Ref(Mov(Close,21,E),-1))</p><p>1</p><p>Now let’s test this trading strategy. Whether you are using BullCharts or</p><p>MetaStock, the testing occurs Over a two-step process, as follows:</p><p>First, the exploration process is run, which in turn identifies all</p><p>occurrences that meet the entry and exit criteria with all stocks loaded</p><p>for testing. Let’s say you have loaded the top 300 stocks for testing and</p><p>load 1500 days of data. The exploration process identifies every set-up</p><p>in price action that meets the specified criteria for all 300 stocks, then</p><p>transfers this data to TradeSim.</p><p>TradeSim then simulates the trading process using the portfolio of 300</p><p>stocks by entering and exiting in accordance with your trading</p><p>conditions and available capital at the time.</p><p>Before you launch the simulation process, you need to define how you intend</p><p>to trade the technique in real time — are you going to enter near the close of</p><p>trade on the day of the signal or will you enter tomorrow morning? Say you</p><p>decide to use end of day data. This requires you to act the next morning,</p><p>meaning you need to delay the entry and exit points by one bar. In order for</p><p>this approach to be reflected in the testing process, you need to include the</p><p>following lines of code:</p><p>ExtFml("TradeSim.EnableDelayOfEntryByOneBar");</p><p>ExtFml("TradeSim.EnableDelayOfAllExitsByOneBar");</p><p>Before starting, you should also ensure that sufficient data is loaded and the</p><p>correct time frame is selected. Once you are satisfied that the testing process</p><p>accurately resembles your approach, you can run the exploration process in</p><p>BullCharts or MetaStock in order to collect the necessary data for TradeSim</p><p>to perform the simulation process.</p><p>MetaStock users need to insert the necessary coding into the explorer while</p><p>making sure that the external formula functions are included in order to</p><p>transfer the data to TradeSim. BullChart users simply write the formula in the</p><p>same manner as for an exploration using either BullScript or the indicator</p><p>menu while using the TradeSim function in the ‘BullScan Manager’. I have</p><p>only included a portion of the complete coding required to activate TradeSim</p><p>and the data transfer process in MetaStock.</p><p>The following parameters directly affect the eventual outcome:</p><p>Range:=((CLOSE-Ref(CLOSE,-1))/Ref(CLOSE,-1))*100;</p><p>EntryTrigger := (Close>Mov(Close,21,E)) AND (Ref(Close,-1)</p><p><=Ref(Mov(Close,21,E)) AND Close>Mov(Close,125,E) AND Range>=5;</p><p>The entry criteria are inserted into the ‘EntryTrigger’ function.</p><p>Let’s say you use end of day data and intend to enter the following</p><p>morning, once the market settles. The nearest price you can reference relative</p><p>to a probable purchase price is the opening, so you must stipulate ‘OPEN’ in</p><p>the ‘EntryPrice’ point, as below:</p><p>EntryPrice := OPEN;</p><p>Your ‘stroke of brilliance’ involving the use of a moving average as an exit</p><p>signal is included in the ‘ExitTrigger’ utility, as follows:</p><p>ExitTrigger := (CLOSE<Mov(Close,21,E)) AND (Ref(Close,-1)</p><p>>=Ref(Mov(Close,21,E),-1));</p><p>As you are trading on end of day data, you should also stipulate that the</p><p>testing process references the opening price, as per the entry, as shown</p><p>below:</p><p>ExitPrice := OPEN;</p><p>Some traders just love to destroy system efficiency by applying tight initial</p><p>trailing stops — followed by much back-patting for disciplined trading.</p><p>However, this is a pitfall to be avoided. Choosing to test without an initial</p><p>stop in place also exposes you to a worst-case scenario that would not be</p><p>encountered if an initial stop was applied under appropriate circumstances.</p><p>When testing a strategy, you should insert a value of zero for your initial</p><p>stop, as follows:</p><p>InitialStop:=0;</p><p>Capital allocation and complex coding</p><p>All of these parameters are determined during the exploration process. Once</p><p>the exploration process is complete, you need to think about capital</p><p>allocation. It really is surprising just how many people I encounter who test</p><p>in a manner that is not even closely related to how they trade — for example,</p><p>they will use $100 000 for testing purposes but trade with $25 000. When</p><p>questioned about the discrepancy, the reply is usually along the lines of, ‘Oh</p><p>the results are more consistent’. This is a classic case of gilding the lily — as</p><p>they don’t have $100K to trade with. If you only have an account balance of</p><p>$25 000 and are only prepared to open five positions at a time, then this is</p><p>how you test. If the results stink, too bad — your approach stinks, simple as</p><p>that. Changing air fresheners does not remove the source of the stench.</p><p>The same applies to coding. I have seen some amazing coding efforts that</p><p>were a credit to the creator’s coding capabilities and that produced on-paper</p><p>results that were equally as impressive. However, the coding was so complex</p><p>that, from my perspective, it was impractical for real-time application. It was</p><p>simply impossible to understand what the formula actually represented or</p><p>what it was attempting to evaluate. While it’s fine to play with exotic</p><p>concepts, for simple application in real time, coding needs to reflect your</p><p>approach and, more importantly, your understanding and trading personality.</p><p>If you do not understand what is being evaluated, you are unlikely to apply</p><p>the concept in real time. I predominately trade on new highs and reversals, so</p><p>I test the identical criteria that I look for in real time. I never add anything to</p><p>enhance the result that I will not use in real time. TradeSim cannot protect us</p><p>from ourselves and the urge to gild the lily by producing superior yet</p><p>artificial results. It’s nice to play, but mixing work with play in the market is</p><p>usually fatal.</p><p>Setting the parameters</p><p>The following trade parameters were used to test the above trading strategy:</p><p>initial capital — $25 000</p><p>portfolio limit — 100 per cent</p><p>maximum number of open positions — 15</p><p>position size model — equal dollar units</p><p>trade size ($ value) — $5000</p><p>pyramid profits — yes</p><p>transaction cost (trade entry) — $20</p><p>transaction cost (trade exit) — $20</p><p>margin requirement — 100 per cent.</p><p>Where:</p><p>Initial capital is the total amount of capital allocated for testing purposes.</p><p>All simulations commence with an opening balance of $25 000.</p><p>Portfolio limit is 100 per cent. In other words, any available capital will</p><p>be allocated should an opportunity arise.</p><p>Maximum number of open positions is limited to 15. This is how many</p><p>trades will be open at any one time. Most people will have somewhere</p><p>between 10 and 20 positions open, so the middle ground was taken.</p><p>Position size model will allocate equal amounts of capital to each trade.</p><p>Trade size is $5000 for each trade.</p><p>Pyramid profits is selected. This means the profits will be left in the</p><p>account and then reinvested once sufficient surplus capital exists. This is</p><p>similar to compounding returns and is a relatively common approach to</p><p>trading the market for many.</p><p>Transaction cost is set at $20 brokerage each way. This will produce a</p><p>more realistic result. Brokerage can impact significantly on strategy</p><p>effectiveness.</p><p>Margin requirement to open a position is set at 100 per cent of trading</p><p>capital, as the strategy is trading shares. That is, 100 per cent of the</p><p>transaction costs must first be cOvered in order to trade the security.</p><p>With regards to the final parameter above, margin requirement, leveraged</p><p>products such as CFDs usually carry a leverage ratio of 10:1 — that is, the</p><p>trader puts up 10 per cent of the required capital to open the position. Hence,</p><p>this value can be adjusted in accordance with your leverage obligation. If you</p><p>want to watch a strategy go through the roof</p><p>traders actually know how a divergence unfolds. This stems</p><p>from not having an understanding of how an indicator is constructed and,</p><p>more importantly, how it relates to price action.</p><p>A divergence occurs when the price channel and price action begin to</p><p>separate, usually during trending conditions. A convergence occurs when the</p><p>price channel and price action begin to gravitate toward each other, usually</p><p>during trending conditions.</p><p>Figure 9.12 shows the change in the closing price that typically occurs</p><p>during bearish and bullish divergences.</p><p>Figure 9.12: closing price during bearish and bullish divergences</p><p>A bearish divergence occurs when a bullish trend continues to rise but the</p><p>closing price is moving closer to the low in the day’s trading range. While</p><p>this gives the Overall impression of rising price action, the indicator shows</p><p>that the range favouring up days is being Overtaken by the range associated</p><p>with down days.</p><p>A bullish divergence occurs when a bearish trend continues to decline but</p><p>the closing price is moving closer to the highs in the day’s trading range.</p><p>While this gives the Overall impression of declining price action, the</p><p>indicator shows that the range favouring down days is being Overtaken by</p><p>the range associated with up days.</p><p>A true bullish divergence occurs when trending price action of a bearish</p><p>nature is offset by a rising index.</p><p>A true bearish divergence occurs when trending price action of a bullish</p><p>nature is offset by a declining index.</p><p>The key to a strong divergence is the conflict in opinion between price and</p><p>indicator development. If a trend is rising and an indicator is declining, the</p><p>rally must be losing support, as the closing price is consistently developing</p><p>nearer to the low. I acknowledge that many variations of divergence activity</p><p>exist; however, you need to be extremely careful once you move beyond the</p><p>above conditions in real time.</p><p>The most dangerous and potentially misleading example that I have seen</p><p>for a divergence is static price action and a trending indicator heading for a</p><p>neutral value. What must any indicator do when price action pauses? By</p><p>design it must return to a more neutral value. A divergence that occurs in this</p><p>situation is not a strong divergence but rather a simple reflection of an easing</p><p>in trend development. A divergence should lead trend development, not</p><p>follow. Added to this, to suggest that an indicator will continue to produce an</p><p>extreme value during non-trending activity is ludicrous.</p><p>Conventional divergence behaviour is not difficult to analyse once you</p><p>understand what the indicator at the bottom of the screen represents —</p><p>usually the closing price in some form. The problem with a more</p><p>conventional approach is that it tends to be very limited in the way it defines</p><p>divergence behaviour. The accepted view is that while divergence behaviour</p><p>can be a strong signal in trending activity, it is not a timing tool. In other</p><p>words, traders may know that there is a divergence is progress, but they</p><p>cannot know when trend direction will change. From a conventional</p><p>approach, I have no problems with accepting this as a logical conclusion —</p><p>considering that price action and the indicator are displayed from alternative</p><p>perspectives. However, when price channels are applied, you have the ability</p><p>to identify a probable point and time of directional change.</p><p>Some of you may find this hard to comprehend; however, you first need to</p><p>understand how the relationship has changed. The adaptation of an indicator</p><p>to price action has altered the manner in which convergence/divergence</p><p>appears.</p><p>When it comes to price channels, there are two types of convergence or</p><p>divergence activity, as follows:</p><p>Price convergence/divergence — which occurs when price action</p><p>changes direction but the price channel continues as previously.</p><p>Channel convergence/divergence — which occurs when price channel</p><p>changes direction but the price action continues as previously.</p><p>I will look at convergences and divergences in turn.</p><p>Convergences</p><p>Figure 9.13 shows examples of convergences.</p><p>Figure 9.13: relative price channel and convergence activity</p><p>Now it is time to lift your understanding up a level. Prepare for your</p><p>understanding of convergences and divergences, and general indicator</p><p>analysis for that matter, to be challenged — sorry. Try to remain as objective</p><p>as possible by reviewing each section on its merits. You need to have an open</p><p>mind if you are going to genuinely appreciate what is before you. Here we</p><p>go.</p><p>What could you tell me about the underlined indicator behaviour at the</p><p>bottom of the chart and the price immediately above that is also underlined</p><p>by the broken lines? Generally speaking, they are the same. The indicator at</p><p>the bottom of the chart reflects the general behaviour of price action Over the</p><p>same period.</p><p>If the indicator and price action are effectively one and the same, the true</p><p>gauge of strength is defined by the Overbought and Oversold values and not</p><p>price action or the indicator value. As the price channel actually defines</p><p>strength, it is logical to conclude that the channel also defines true</p><p>convergence or divergence behaviour. Price action is effectively the axis on</p><p>which the price channel oscillates; therefore, divergence analysis is defined</p><p>by channel behaviour and not price action. If this is applicable to price action</p><p>and the relative price channel, this is also applicable to the conventional</p><p>indicator at the bottom of the chart. I’ll bet this line of thought has your head</p><p>spinning, because it flies in the face of everything that you have ever been</p><p>told about divergence analysis. I will try to explain this is in slightly more</p><p>detail by referring to figure 9.13.</p><p>In my discussion of figure 9.13, I will first look at the conventional</p><p>indicator, then at the price channel.</p><p>At point ‘A’ on the chart, the conventional indicator has generally drifted in</p><p>a non-trending direction. It gravitated around the values of 45 per cent and 55</p><p>per cent without attempting to break in either direction. In early March 2004,</p><p>the relative channel index thrusts upward. Conventional thinking would</p><p>indicate that the stock is gaining in strength; however, this has not been</p><p>confirmed. Remember that the 70 per cent value on the conventional</p><p>indicator is relative to price action. For charting purposes it is fixed at a</p><p>predetermined position, giving the impression of having a static value. What</p><p>traders cannot know is whether the true value of the Overbought region is</p><p>declining or whether there is the presence of rising strength. They assume</p><p>that strength is rising purely because the indicator value is rising.</p><p>Remember</p><p>The values of Overbought and Oversold regions change in unison with trend development.</p><p>Traders assume the value is static for no other reason than that it is a fixed point of</p><p>reference on the chart.</p><p>With regards to point A on the chart and the price channel, while price action</p><p>has been rising since November 2003, the price channel has continued in a</p><p>non-trending manner. Remember — the price channel reflects the bias</p><p>associated with the closing price. In order to help you develop a mental</p><p>picture of what is before you, try to remember that the price channels ebb and</p><p>flow around price action according to the bias that exists as the trend</p><p>develops. In this case, price action is rising but the channel moves in a</p><p>manner contrary to price action. This is most obvious if you look at the</p><p>formation of the lower section of the price channel. Rather than indicating the</p><p>presence of an increasing bullish bias, as suggested by the conventional</p><p>indicator at the bottom of the chart, there is actually a last-minute dip in</p><p>strength that draws the channel downward. This is what gives the false</p><p>impression of increasing strength with the conventional indicator.</p><p>Remember</p><p>It’s the bias that exists between higher and lower closes that creates the price channel.</p><p>Now let’s look at point ‘B’ on the chart, starting with the conventional</p><p>indicator. In mid-September the relative channel</p><p>or turn unbelievably pear-</p><p>shaped, use this function. It’s a real wake-up call with regard to some</p><p>accepted techniques.</p><p>The above parameters will also be used for all following testing and</p><p>development purposes throughout the book.</p><p>Figure 1.2 shows the TradeSim screen where these parameters can be set.</p><p>Figure 1.2: simulation account parameters — TradeSim</p><p>Having decided that the trading parameters reflect the required application,</p><p>it’s now time to test the strategy.</p><p>Testing the strategy</p><p>As mentioned, TradeSim trades in the identical manner to real-time</p><p>application in that, once there is full capital allocation, all further trades are</p><p>ignored until an existing trade is closed. This is carried through the entire</p><p>testing process. The end result highlights whether the tested strategy is a</p><p>stroke of brilliance or another mindless idea taking up valuable time.</p><p>Table 1.1 shows the results produced by TradeSim.</p><p>Table 1.1: moving average crossOver</p><p>Profit summary</p><p>Profit status PROFITABLE</p><p>Starting capital $25 000.00</p><p>Finishing capital $62 322.81</p><p>Maximum equity/(date) $39 284.77 (31/03/2005)</p><p>Minimum equity/(date) −$8419.12 (18/10/2000)</p><p>Gross trade profit $145 755.74 (583.02%)</p><p>Gross trade loss −$108 432.93 (−433.73%)</p><p>Total net profit $37 322.81 (149.29%)</p><p>Average profit per trade $65.71</p><p>Profit factor 1.3442</p><p>Profit index 25.61%</p><p>Total transaction cost $22 720.00</p><p>Total slippage $0.00</p><p>Daily compound interest rate 0.0447%</p><p>Annualised compound interest rate 17.6986%</p><p>Trades processed 1676</p><p>Trades taken 568</p><p>Partial trades taken 0</p><p>Trades rejected 1108</p><p>Winning trades 161 (28.35%)</p><p>Losing trades 407 (71.65%)</p><p>Largest winning trade/(date) $12 959.48 (10/12/2004)</p><p>Largest losing trade/(date) −$1905.63 (31/05/2001)</p><p>Average winning trade $905.32</p><p>Average losing trade −$266.42</p><p>Average win/average loss 3.3981</p><p>Trade duration statistics (all trades)</p><p>Maximum trade duration 107 (days)</p><p>Minimum trade duration 1 (day)</p><p>Average trade duration 16 (days)</p><p>Winning trades</p><p>Maximum trade duration 107 (days)</p><p>Minimum trade duration 1 (day)</p><p>Average trade duration 35 (days)</p><p>Losing trades</p><p>Maximum trade duration 63 (days)</p><p>Minimum trade duration 1 (day)</p><p>Average trade duration 8 (days)</p><p>Consecutive trade statistics</p><p>Maximum consecutive winning trades 7</p><p>Maximum consecutive losing trades 23</p><p>Average consecutive winning trades 1.48</p><p>Average consecutive losing trades 3.77</p><p>Relative drawdown</p><p>Maximum dollar drawdown/(date) $5604.45 (3/04/2003)</p><p>Maximum percentage drawdown/(date) 19.9900% (10/04/2001)</p><p>Absolute (peak to valley) dollar drawdown</p><p>Maximum dollar drawdown $12 496.32 (43.0400%)</p><p>Capital peak/(date) $29 037.20 (13/03/2000)</p><p>Capital valley/(date) $16 540.88 (18/10/2000)</p><p>The results shown in table 1.1 clearly define how the tested strategy</p><p>performed Over the testing period. The annualised return was acceptable at</p><p>17.69 per cent; however, a success rate of 28.35 per cent can only be</p><p>described as dreadful. Commencing with an account balance of $25 000 the</p><p>strategy finished with $62 322.81, thereby producing a net profit of $37</p><p>322.81 or 149.29 per cent.</p><p>Testing results to focus on</p><p>While it is not essential to pour Over the most intricate detail and to review</p><p>your strategy from every conceivable angle — although TradeSim will</p><p>certainly accommodate you from this perspective — you need to ensure that</p><p>the strategy does enjoy a level of consistency if your intention is to genuinely</p><p>trade it in real time. I tend to focus on the following:</p><p>consistency of equity curve</p><p>consistency of yearly return</p><p>relative drawdown — this should not be excessive and should</p><p>consistently be less than the peak to valley drawdown</p><p>trades manager — anything extraordinary needs to be removed</p><p>average win — there should be a healthy difference between average</p><p>win and average loss</p><p>average loss — this should be less than capital at risk</p><p>winning trades at greater than 50 per cent</p><p>consecutive winning trades that outnumber consecutive losses</p><p>consecutive losing trades — if these are dominant, they should not be</p><p>excessive</p><p>trades processed — these should not be Overly high as too many</p><p>opportunities make it difficult to replicate the average performance in</p><p>real time.</p><p>Some of the factors will be discussed below. I do not become too analytical</p><p>when scrutinising the results. I’m looking for a strategy that is consistent and</p><p>enjoys an acceptable rate of success — preferably around 50 per cent — with</p><p>an average drawdown that is manageable. While I do not like to see the</p><p>relative drawdown push beyond 10 per cent, this is gOverned by the strategy</p><p>in place and whether it targets rally or position trading — naturally strategies</p><p>that absorb retracement activity will experience higher drawdowns due to a</p><p>wider ranging stop.</p><p>Factors that may be Overlooked</p><p>When reviewing any strategy under development, try to avoid the urge to</p><p>gloss Over the negative impact — this is what will ultimately destroy capital</p><p>when in application. The following pieces of information are commonly</p><p>Overlooked when evaluating a strategy.</p><p>Peak to valley drawdown</p><p>This is the worst-case scenario, or the largest drop recorded from a capital</p><p>high to a capital low. TradeSim searches for the absolute high in equity value</p><p>minus the absolute low in equity value. It then displays this as the absolute</p><p>possible drawdown. While this is the absolute worst possible result, this is</p><p>not to say that this particular simulation experienced this degree of</p><p>drawdown. A wide range between the ‘Relative drawdown’ and the</p><p>‘Absolute peak to valley drawdown’ tends to suggest that the trailing stop</p><p>within the strategy is working to a degree.</p><p>You cannot determine future drawdown nor can you prevent it entirely.</p><p>Scrapping a system on the peak to valley drawdown alone is not logical.</p><p>While it is an important piece of the puzzle, it is only one piece. You should</p><p>have risk management techniques in place and capital allocation rules that</p><p>aim to limit the exposure to adverse activity.</p><p>Relative drawdown</p><p>Relative drawdown is the impact of a series of losses. Occasionally, losses</p><p>outweigh gains causing a drawdown on capital. This is a natural part of</p><p>trading, so it should never be Overlooked. The tested strategy experienced a</p><p>losing streak of $5604.45. The worst-case scenario — that is, the possible</p><p>peak to valley drawdown — was not experienced with this particular</p><p>simulation. Accepted market practice suggests that a strategy with a relative</p><p>drawdown of less that 20 per cent is acceptable. I consider that this is a</p><p>somewhat misleading conclusion as the drawdown must be relative to the</p><p>strategy.</p><p>Average loss</p><p>Losses range from just shy of break-even to outright disastrous. These are</p><p>usually referred to as minimum and absolute, with most losses somewhere in</p><p>between. While both values are easy to identify, the problem is that traders</p><p>will seldom experience either, meaning defining expected loss with a degree</p><p>of consistency can become a difficult process. While you can speculate about</p><p>the range of losses to be encountered, the only effective value, in my opinion,</p><p>is average loss. If you know that on average you can expect to lose X amount</p><p>of dollars by the time your trailing stop is triggered, you have a means for</p><p>defining and, more importantly, managing risk. The average loss is the</p><p>amount that you can expect to lose, all things being equal Over the longer</p><p>term. Some trades will lose less while others will go beyond average.</p><p>Personally, I prefer to focus on averages and relative values rather than</p><p>absolutes. It is surprising just how many traders focus primarily on absolute</p><p>values when in reality they are seldom experienced. Providing my technique</p><p>performs within the accepted boundaries of average behaviour, I am happy</p><p>that everything is performing as expected. Only when I begin to experience</p><p>results that go beyond average do I become concerned.</p><p>Tip</p><p>Is your average loss consistent with your position risk? Say you trade with a $25 000</p><p>account and you are going to apply the 2 per cent rule of risk exposure — that is, you are</p><p>only prepared to risk $500 in each trade. With</p><p>the strategy tested above, the average loss on</p><p>a $5000 position was only $266.42, meaning you would be well under your maximum level</p><p>of exposure. You could trade this particular technique knowing that on average your losses</p><p>would be less than your 2 per cent limit. This in turn directly affects your long-term</p><p>survival. This is one of the reasons why you should always test using relative values as it</p><p>gives the results added meaning. Sure, the net profit may not look that impressive, but at</p><p>least it’s realistic and, more importantly, relative to your actions.</p><p>Potential trades</p><p>Another area of concern is the number of potential trades identified during</p><p>the testing process. If you experience a reduced number of potential trades,</p><p>the likelihood of replicating your testing results improves significantly, as</p><p>opposed to a strategy that produces many signals. The higher the number of</p><p>trading opportunities identified, the greater the scope for variation in the</p><p>testing results, and the more difficult it becomes to replicate historical</p><p>performance. This is simply because the benchmark is usually so varied,</p><p>which leads to an almost infinite number of potential outcomes.</p><p>Anomalies</p><p>Once the testing process is complete, you should have a look at the Trade</p><p>Database Manager screen, shown in figure 1.3. Occasionally, a one-off trade</p><p>will completely distort the Overall results — for example, I was developing a</p><p>volume-based strategy for personal use and suddenly the results went through</p><p>the roof. I looked at the results and thought, ‘Before I get too excited, let’s</p><p>see whether it’s a bonza technique or whether I have a one-off trade that has</p><p>produced a phenomenal return’. Through looking at the positions taken by</p><p>TradeSim within the Trade Database Manager screen, as shown in figure 1.3,</p><p>I found two trades that were out of character. PDN produced a 5794.74 per</p><p>cent profit while SDI produced a return of 2557.58 per cent. No wonder my</p><p>results where looking good. TradeSim allowed me to remove these two</p><p>offending stocks from the testing process.</p><p>Figure 1.3: Trade Database Manager</p><p>In order to do away with any undesirable stock from the testing process,</p><p>simply remove the tick from the trade box to the left in the ‘Trade’ column.</p><p>As there was a reasonable number of stocks producing results in excess of</p><p>500 per cent, I decided that all of the remaining trading opportunities were</p><p>not out of character with strategy application and so did not need to be</p><p>removed. Having removed PDN and SDI, I had the ability to review the</p><p>technique using more realistic values.</p><p>Tip</p><p>Use the ‘Random Walk’ function when testing as this will allow TradeSim to produce a</p><p>range of results by repeating the simulation process.</p><p>Net return versus consistency of returns</p><p>The most common approach when looking at test results is to go straight to</p><p>the net return and either say, ‘Nah, no good’ or ‘Yep, I can live with that per</p><p>year’. If the return is deemed not good enough, this conclusion is</p><p>immediately followed by swift contact with the delete button. If a conclusion</p><p>in the affirmative is drawn, without delay trading is commenced. Many</p><p>people automatically assume that this return was spread across the entire</p><p>testing period with a degree of consistency; however, this is not always the</p><p>case.</p><p>When looking at net returns, it is also important to look at the equity curve</p><p>and yearly profits generated by the trading simulation in order to put the net</p><p>return into context. The strategy tested above produced a net return of 17 per</p><p>cent per year, which seems pretty good. However, the equity curve and yearly</p><p>profits from the strategy are shown in figures 1.4 and 1.5. If you look at these</p><p>figures, you can see that the strategy has performed in a very ordinary</p><p>manner, with the exception of 2004. If you were trying to trade this particular</p><p>strategy in real time, you would have experienced very spasmodic results.</p><p>The equity curve is less than flattering with the bulk of its activity occurring</p><p>in a non-trending manner. The bulk of profits were generated Over a very</p><p>short period of time, which is not practical for real-time application.</p><p>Figure 1.4: closed equity curve for moving average crossOver strategy</p><p>Figure 1.5: yearly profits for moving average crossOver strategy</p><p>Remember</p><p>The net profit is never as important as the consistency with which the return is generated or</p><p>the risk taken to achieve the return. It’s a real pity that greedy eyes obscure reality. So</p><p>many inexperienced traders focus on the net return when this is only one element of a</p><p>successful strategy.</p><p>Overview of ground rules for testing</p><p>The basic ground rules for testing in a realistic manner are as follows:</p><p>code the strategy as closely as possible to how you intend to trade</p><p>don’t test the strategy in a manner that simply enhances the result</p><p>use the amount of capital you have available</p><p>limit the number of positions to what you will need to follow in real</p><p>time</p><p>include realistic brokerage</p><p>load sufficient data to include a range of conditions.</p><p>Selective testing</p><p>The most common approach to loading data and then testing strategies is to</p><p>load a selected number of periods — say, 1500 periods on daily or 300</p><p>periods on weekly commencing from the most recent date — and then let the</p><p>process run its course. Over time I have found that the development of short</p><p>side trading techniques needs to be directed more precisely toward a bearish</p><p>environment. While this may appear an obvious comment, it is possible to</p><p>produce a positive result with a long side strategy in bearish conditions far</p><p>more easily than the reverse — that is, looking to apply a short side strategy</p><p>in bullish conditions. As the market develops in a different manner when</p><p>declining, a more selective approach to testing is required for this market, as</p><p>opposed to rather than the more generalised approach to market conditions</p><p>that is acceptable when long side testing.</p><p>TradeSim incorporates the following function that allows you to limit your</p><p>testing to a predetermined period in time. If you know that the market started</p><p>to weaken in August 2001 and did not turn bullish again until March 2003</p><p>and you are testing a short side strategy, you can include the following in</p><p>order to limit your testing to this period in time:</p><p>ExtFml( "TradeSim.SetStartRecordDate",1,8,2001);</p><p>ExtFml( "TradeSim.SetStopRecordDate",31,3,2003);</p><p>Some may say that this is curve fitting results to a particular time in the</p><p>market; however, I disagree, although a specific period of market activity is</p><p>being targeted. Short side trading is normally the domain of more</p><p>experienced traders; therefore, they will usually have specific strategies</p><p>tailored for long and short side trading. The inexperienced trader will attempt</p><p>to apply a generic long style strategy without question while giving little or</p><p>no consideration to shorting the market or determining underlying conditions.</p><p>The seasoned trader does not curve fit a strategy simply to enhance the results</p><p>but rather curve fits a strategy to suit current conditions. The results are</p><p>merely a consequence of the trader adapting to the environment.</p><p>∗∗∗</p><p>Mechanical testing cannot replicate real-time trading for a multitude of</p><p>reasons. Programs such as TradeSim are used to refine concepts ready for</p><p>real-time testing, not real-time application. While TradeSim is clinical in its</p><p>application, the average trader is anything but clinical. The one thing</p><p>TradeSim cannot factor in is us.</p><p>This chapter has discussed the use of TradeSim to test trading strategies</p><p>and highlighted the factors within the testing results that are the most</p><p>important to focus on. The following chapters will test trade strategies that</p><p>push the boundaries of accepted practice, using TradeSim to generate test</p><p>results and focusing on the same factors to compare returns and profitability.</p><p>1 Before anyone emails me with more simplified coding — that is,</p><p>Cross(Mov(Close,21,E),Close) — I have a personal preference for a more</p><p>precise approach when identifying set-up conditions. The</p><p>reasons behind</p><p>this preference are a bit longwinded but, in simple terms, if I use this</p><p>approach, I know that the set-up is accurate to how I trade. Occasionally, I</p><p>have produced alternative results using the ‘Cross’ reference, contrary to</p><p>what I would expect. While in theory any variation in the end result using</p><p>the cross function should not occur, I have encountered a number of</p><p>situations where this has happened. As I have been unable to identify the</p><p>source of the problem, I have ceased using the cross function, even though</p><p>it is in theory a simplified process for achieving the same outcome. While</p><p>the variation in results was not significant, it was sufficient to cause</p><p>concern.</p><p>Chapter 2</p><p>Index regression</p><p>What’s relative?</p><p>The most popular indicator in circulation today would have to be the relative</p><p>strength index (RSI). While there is plenty of literature available on its</p><p>interpretation, I will cOver the basics here before building upon its</p><p>application in a way that goes well beyond the conventional approach and</p><p>begins to move into uncharted waters. If we going to apply the RSI in new</p><p>ways (or any indicator for that matter), it helps if we have an understanding</p><p>of what it is attempting to evaluate.</p><p>The original concepts behind the RSI were discussed by J. Welles Wilder</p><p>Jnr in New Concepts in Technical Trading Systems. In this book, Wilder gave</p><p>exacting detail on how the RSI is calculated; however, in order to save you</p><p>the trouble of working it out, I have included the coding here. While this is</p><p>my personal approach for coding the RSI, many similar versions exist and all</p><p>are consistent with the original — so don’t dismiss any alternative approach</p><p>out of hand. While this particular version differs slightly to that in my</p><p>previous example in The Next Step to Share Trading Success, the end result is</p><p>consistent with the original. (As I write my own indices I have amended my</p><p>existing formula to the current version for continuity.) The coding for the RSI</p><p>is as follows:</p><p>Periods:=input("Calculation Periods", 14, 1, 250);</p><p>Range:=(2*Periods)-1;</p><p>UpDay:=ma(If(CLOSE>Ref(CLOSE,-1),CLOSE-Ref(CLOSE,-1),0),Range,E);</p><p>DownDay:=ma(If(CLOSE<Ref(CLOSE,-1),Ref(CLOSE,-1)-CLOSE,0),Range,E);</p><p>100-(100/(1+(UpDay/(DownDay+0.000001))))</p><p>If you look closely at the above formula, you can see that it is not Overly</p><p>complicated. The formula merely creates two moving averages — one</p><p>depicting the change behind higher closing prices and one reflecting the</p><p>change associated with lower closing prices. In other words, the formula:</p><p>creates a moving average of the change between today’s closing price</p><p>and yesterday’s closing price when today’s closing price is higher than</p><p>yesterday’s</p><p>creates a separate moving average of the range between today’s closing</p><p>price and yesterday’s closing price when today’s closing price is lower</p><p>than yesterday’s</p><p>defines the relationship between the two moving averages as a</p><p>percentage.</p><p>The RSI is only capable of recording one value per week. If a week produces</p><p>a higher closing price, the range between last week and this week is</p><p>calculated using the ‘UpDay’ line. As the closing price was higher, the</p><p>‘DownDay’ will record a value of zero.</p><p>Tip</p><p>When using an indicator as confirmation, ask yourself whether it can be applied to price</p><p>action. If it cannot be applied to price action, how do you know whether its parameters are</p><p>relative to trend development? If you are unable to determine whether its parameters are</p><p>relative to trend development, why are you using it?</p><p>Knowing how the relative strength index is constructed is one thing, but</p><p>understanding how this relates to price action is another issue completely.</p><p>According to the more conventional approach to confirmation analysis, the</p><p>RSI is usually analysed from two perspectives — value and behaviour.</p><p>Value</p><p>Value is where extreme values in indicator readings are looked for.</p><p>Extremely high or low values are considered by many to suggest that current</p><p>trend activity is unsustainable — for example:</p><p>a stock is considered ‘Overbought’ when the RSI produces a value</p><p>greater than 70 per cent</p><p>a stock is considered as ‘Oversold’ when the RSI produces a value less</p><p>than 30 per cent.</p><p>The first thing novice traders need to do is scrap this ridiculous notion of</p><p>‘Overbought’ and ‘Oversold’. To categorize these zones in such a way is total</p><p>nonsense when attempting to apply it the broader market. It is also potentially</p><p>misleading, as it can create the impression that stocks are either super cheap</p><p>or too expensive, and so can set a mental trap for your subconscious. Stocks</p><p>can spend extended periods in these ‘Overbought’ or ‘Oversold’ regions and</p><p>this reflects ongoing strength or weakness. In reality, these terms simply</p><p>imply that trend behaviour is potentially becoming unsustainable, meaning it</p><p>is more likely that the trend or rally is nearing its conclusion.</p><p>Behaviour</p><p>The second approach involves behavioural analysis. In this approach,</p><p>behaviour takes precedence Over indicator values. The most obvious aspect</p><p>of behaviour analysis would be a divergence — where, for example, price</p><p>action may continue to rise, but the RSI declines. This means that the closing</p><p>price is finishing lower so the appearance of rising price action stems from</p><p>bullish intra-day activity. Divergences are best described as a difference in</p><p>opinion between price action and indicator values. Put simply, a bullish</p><p>divergence occurs when prices are declining but indicator values are</p><p>increasing and a bearish divergence occurs when prices are increasing but</p><p>indicator values are decreasing. Divergences carry greater significance when</p><p>they commence at extreme RSI values.</p><p>The relative strength index and value analysis</p><p>I will look at value analysis first. On figure 2.1, I have highlighted points in</p><p>indicator behaviour that would be of significance in value analysis.</p><p>Figure 2.1: relative strength index — value analysis</p><p>The areas are as follows:</p><p>1 RSI values higher than 70 per cent are associated with unsustainable</p><p>trend development. It is generally considered that entry should be avoided</p><p>once the RSI moves into the Overbought region. While this appears to be</p><p>a justifiable conclusion to draw when taken at face value, I have included</p><p>a vertical line on the chart where the RSI swings above 70 per cent.</p><p>2 It is common to see many textbook examples highlighting why the RSI</p><p>crossing below 70 per cent is considered as an appropriate time to sell.</p><p>The reality is somewhat different. If you look back to the two occurrences</p><p>labelled ‘1’, in these situations the RSI has dropped below 70 per cent</p><p>from being Overbought; however, on each occasion price action has not</p><p>retraced but merely paused. Inexperienced traders will see a move below</p><p>70 per cent as a confirmation signal to either close a long position or short</p><p>the stock in an attempt to capture any downside movement. While this</p><p>may appear sound in theory, it is a practice that is fraught with danger.</p><p>While the RSI remains above 50 per cent, a bullish dominance underpins</p><p>price action. It is not logical to short stocks that have an underlying</p><p>bullish bias supporting price action.</p><p>3 The area at point three is a very poor attempt at highlighting a failure</p><p>swing, which I will have a closer look at shortly. As shown in figure 2.2, a</p><p>failure swing occurs when an indicator rebounds following a decline,</p><p>usually from an extreme value, but fails to exceed the previous high value</p><p>before collapsing. Failure swings are equally applicable to bullish and</p><p>bearish behaviour, and like divergences they should occur following an</p><p>extreme value within the Overbought or Oversold region.</p><p>Figure 2.2: failure swings</p><p>4 An RSI value of less than 30 per cent is considered to suggest that the</p><p>current downtrend is potentially becoming unsustainable, meaning you</p><p>should hold off if you were intending to short the stock. In reality, it</p><p>merely reflects a strong bias in favour of bearish activity. Due to the</p><p>slightly different and often more aggressive nature in how down trending</p><p>activity develops,</p><p>an extremely low RSI value may well coincide with</p><p>trend conclusion as trend exhaustion can occur quickly; however, this is</p><p>not always the case once a downtrend becomes established. Some stocks</p><p>do develop on the back of strong bearish RSI values.</p><p>5 The bullish move above the 30 per cent barrier is usually perceived to</p><p>be a signal to close any short positions. Again, many inexperienced</p><p>traders will also view this occurrence as a confirmation signal for long</p><p>side trading. Indeed, this approach is strongly advocated by many</p><p>traditionalists as it suggests that the stock is gaining in strength. While</p><p>this is a logical conclusion to draw at face value, the dominating factor</p><p>that many Overlook is that the stock is still predominately bearish. Rather</p><p>than viewing a stock that has an RSI value that is less than 50 per cent and</p><p>rising as gaining in strength, this should be seen as an easing in weakness.</p><p>Strength only becomes the dominant element once the RSI is greater than</p><p>50 per cent. It is surprising the lengths many people will go to, so they can</p><p>‘gild the lily’ in order to produce the desired outcome.</p><p>As will be discussed later in the book, down trending stocks do not require</p><p>strong bearish RSI values in order to remain sustainable in the same manner</p><p>as uptrends require strong bullish values in order to survive.</p><p>Looking purely at indicator values, there are only two areas of major</p><p>interest when using a more conventional approach. Extremely high or low</p><p>values are considered to suggest that the current trend is becoming</p><p>unsustainable and moves by the RSI either below 70 per cent or above 30 per</p><p>cent suggest that there is a potential trend change in progress. In reality,</p><p>extreme values show very little until after the fact, which can be of little</p><p>benefit in real time. What is of even bigger concern is that, because</p><p>conventional analysis only gives two main areas of interest, the majority of</p><p>people fail to utilise more than 1 per cent of the available information hidden</p><p>within the relative strength index.</p><p>The relative strength index and behavioural</p><p>analysis</p><p>Behavioural analysis involves analysing the relationship that exists between</p><p>the indicator and price, or analysing the indicator itself. Naturally, traders are</p><p>always on the lookout for divergence behaviour between the stock and</p><p>indicator; however, you can also look for double tops, head and shoulders,</p><p>wedges and so on that are forming within the indicator itself. Try to</p><p>remember with bearish divergences that the indicator is reflecting a declining</p><p>value of the closing price. Likewise, the opposite is true for a bullish</p><p>divergence — the indicator is reflecting a higher close while the bulk of intra-</p><p>day activity declines.</p><p>On figure 2.3 I have again highlighted points that would be of significance</p><p>in behavioural analysis. The areas are as follows:</p><p>Figure 2.3: relative strength index — behavioural analysis</p><p>1 Behaviour has the potential to tell you far more than just a percentage</p><p>value alone. As discussed, many people believe that an RSI value greater</p><p>than 70 per cent means the trend is potentially becoming unsustainable;</p><p>however, I disagree with this notion under certain conditions. When a new</p><p>trend initially commences, it is relatively common for range bound</p><p>indicators to produce extremely high values. I view this as extreme bullish</p><p>enthusiasm, not bullish exhaustion. Once a trend is underway, only then</p><p>am I prepared to accept that extremely high values are potential signs of</p><p>exhaustion.</p><p>The reason why I will not automatically accept that 70 per cent is an</p><p>Overbought signal early in trend development is due to the behaviour of</p><p>the closing price. The upward swing of the closing price dramatically</p><p>shifts the balance in the RSI calculation process. When a trend initially</p><p>springs into life, it is not uncommon for the stock to experience a run of</p><p>higher closing prices. The market wants to enter this stock while prices</p><p>are relatively low. In the James Hardie Industries (JHX) chart, shown in</p><p>figure 2.3, there is only one week where the closing price finished lower</p><p>than the previous week. If you think back to how the RSI calculates its</p><p>value, this means there was only one week that produced a bearish value</p><p>while there were nine weeks that produced a bullish value. In other words,</p><p>there were eight weeks that produced a value of zero. This imbalance</p><p>between bullish and bearish behaviour has a significant impact on the RSI</p><p>calculation process, which in turn causes the indicator to rapidly produce</p><p>an extremely bullish value simply because the calculation process lacks</p><p>similar bearish input Over the same period.</p><p>2 Divergences can be a strong signal when they are correctly identified.</p><p>While there is a range of possible divergence relationships between price</p><p>action and the indicator, the only true divergence in my opinion is the one</p><p>that has developed here with JHX. While price action is rising, the RSI is</p><p>producing a series of lower values commencing from an exceptionally</p><p>high value. This identifies the fact that while there is the visual evidence</p><p>of rising prices with higher highs and higher lows, the closing price is</p><p>consistently decreasing in value by developing nearer to the low in price</p><p>action as the trend matures. While intra-day activity has the ability to push</p><p>prices higher, the professional end of the market is not prepared to</p><p>continue supporting bullish activity. The divergence shown is very short</p><p>term in development, so its impact should also be fairly brief. The impact</p><p>of divergences on trend development is no different from that of candles</p><p>and trading patterns — the shorter the pattern is from start to maturity, the</p><p>less the impact will potentially be Over longer term trend development.</p><p>As mentioned, it is also preferable to look for a divergence to start within</p><p>the Overbought or Oversold regions. Divergences that develop mid-range</p><p>in indicator value are usually of little or no significance.</p><p>3 While my initial intention with figure 2.3 was to highlight the</p><p>relationship between price action and the RSI at points one and two, I also</p><p>noticed a more clearly definable failure swing that, while it could have</p><p>caught the majority in real time, most would perceive as nothing more</p><p>than a bump in indicator development now. The RSI rebounds following a</p><p>recent decline in value, but it fails to test the previous high. Some traders</p><p>strongly advocate the use of failure swings as a signal of continued</p><p>decline; however, I consider that this type of behaviour is generally</p><p>useless in real time. Although I’m the first to acknowledge that failure</p><p>swings will coincide with a continued decline, this style of analysis works</p><p>best to the left of screen. The only way you can confirm that a pattern is a</p><p>genuine failure swing is through the failure of the upward swing itself.</p><p>This can only be known with any certainty several periods after the fact,</p><p>once the RSI has produced a second but lower peak and is declining</p><p>notably.</p><p>Exercise</p><p>Starting from early July 2005 on figure 2.3, slowly slide a sheet of paper from left to right.</p><p>Can you see the divergence developing as the rally matures? Now slide the paper right until</p><p>you get to the peak of the failure swing, as indicated by the label ‘3’. Slide the paper very</p><p>slowly to the right and watch how the indicator declines. It would be possible to extend a</p><p>declining line of resistance downward, as I have indicated with the broken line. In real</p><p>time, many would consider a true failure swing to be in progress. In late September the RSI</p><p>kicks upward and suddenly the failure swing has failed as the declining line of resistance</p><p>on the RSI has been comfortably penetrated by bullish activity. With a value pushing 60</p><p>per cent there would be little reason to suspect otherwise in real time — please note the</p><p>continued reference to ‘real time’.</p><p>The point that I am trying to make is that the appearance of the failure swing</p><p>changed as price action evolved. In hindsight, this would not be considered as</p><p>a prominent failure swing; however,</p>traders actually know how a divergence unfolds. This stems
from not having an understanding of how an indicator is constructed and,
more importantly, how it relates to price action.
A divergence occurs when the price channel and price action begin to
separate, usually during trending conditions. A convergence occurs when the
price channel and price action begin to gravitate toward each other, usually
during trending conditions.
Figure 9.12 shows the change in the closing price that typically occurs
during bearish and bullish divergences.
Figure 9.12: closing price during bearish and bullish divergences
A bearish divergence occurs when a bullish trend continues to rise but the
closing price is moving closer to the low in the day’s trading range. While
this gives the Overall impression of rising price action, the indicator shows
that the range favouring up days is being Overtaken by the range associated
with down days.
A bullish divergence occurs when a bearish trend continues to decline but
the closing price is moving closer to the highs in the day’s trading range.
While this gives the Overall impression of declining price action, the
indicator shows that the range favouring down days is being Overtaken by
the range associated with up days.
A true bullish divergence occurs when trending price action of a bearish
nature is offset by a rising index.
A true bearish divergence occurs when trending price action of a bullish
nature is offset by a declining index.
The key to a strong divergence is the conflict in opinion between price and
indicator development. If a trend is rising and an indicator is declining, the
rally must be losing support, as the closing price is consistently developing
nearer to the low. I acknowledge that many variations of divergence activity
exist; however, you need to be extremely careful once you move beyond the
above conditions in real time.
The most dangerous and potentially misleading example that I have seen
for a divergence is static price action and a trending indicator heading for a
neutral value. What must any indicator do when price action pauses? By
design it must return to a more neutral value. A divergence that occurs in this
situation is not a strong divergence but rather a simple reflection of an easing
in trend development. A divergence should lead trend development, not
follow. Added to this, to suggest that an indicator will continue to produce an
extreme value during non-trending activity is ludicrous.
Conventional divergence behaviour is not difficult to analyse once you
understand what the indicator at the bottom of the screen represents —
usually the closing price in some form. The problem with a more
conventional approach is that it tends to be very limited in the way it defines
divergence behaviour. The accepted view is that while divergence behaviour
can be a strong signal in trending activity, it is not a timing tool. In other
words, traders may know that there is a divergence is progress, but they
cannot know when trend direction will change. From a conventional
approach, I have no problems with accepting this as a logical conclusion —
considering that price action and the indicator are displayed from alternative
perspectives. However, when price channels are applied, you have the ability
to identify a probable point and time of directional change.
Some of you may find this hard to comprehend; however, you first need to
understand how the relationship has changed. The adaptation of an indicator
to price action has altered the manner in which convergence/divergence
appears.
When it comes to price channels, there are two types of convergence or
divergence activity, as follows:
Price convergence/divergence — which occurs when price action
changes direction but the price channel continues as previously.
Channel convergence/divergence — which occurs when price channel
changes direction but the price action continues as previously.
I will look at convergences and divergences in turn.
Convergences
Figure 9.13 shows examples of convergences.
Figure 9.13: relative price channel and convergence activity
Now it is time to lift your understanding up a level. Prepare for your
understanding of convergences and divergences, and general indicator
analysis for that matter, to be challenged — sorry. Try to remain as objective
as possible by reviewing each section on its merits. You need to have an open
mind if you are going to genuinely appreciate what is before you. Here we
go.
What could you tell me about the underlined indicator behaviour at the
bottom of the chart and the price immediately above that is also underlined
by the broken lines? Generally speaking, they are the same. The indicator at
the bottom of the chart reflects the general behaviour of price action Over the
same period.
If the indicator and price action are effectively one and the same, the true
gauge of strength is defined by the Overbought and Oversold values and not
price action or the indicator value. As the price channel actually defines
strength, it is logical to conclude that the channel also defines true
convergence or divergence behaviour. Price action is effectively the axis on
which the price channel oscillates; therefore, divergence analysis is defined
by channel behaviour and not price action. If this is applicable to price action
and the relative price channel, this is also applicable to the conventional
indicator at the bottom of the chart. I’ll bet this line of thought has your head
spinning, because it flies in the face of everything that you have ever been
told about divergence analysis. I will try to explain this is in slightly more
detail by referring to figure 9.13.
In my discussion of figure 9.13, I will first look at the conventional
indicator, then at the price channel.
At point ‘A’ on the chart, the conventional indicator has generally drifted in
a non-trending direction. It gravitated around the values of 45 per cent and 55
per cent without attempting to break in either direction. In early March 2004,
the relative channel index thrusts upward. Conventional thinking would
indicate that the stock is gaining in strength; however, this has not been
confirmed. Remember that the 70 per cent value on the conventional
indicator is relative to price action. For charting purposes it is fixed at a
predetermined position, giving the impression of having a static value. What
traders cannot know is whether the true value of the Overbought region is
declining or whether there is the presence of rising strength. They assume
that strength is rising purely because the indicator value is rising.
Remember
The values of Overbought and Oversold regions change in unison with trend development.
Traders assume the value is static for no other reason than that it is a fixed point of
reference on the chart.
With regards to point A on the chart and the price channel, while price action
has been rising since November 2003, the price channel has continued in a
non-trending manner. Remember — the price channel reflects the bias
associated with the closing price. In order to help you develop a mental
picture of what is before you, try to remember that the price channels ebb and
flow around price action according to the bias that exists as the trend
develops. In this case, price action is rising but the channel moves in a
manner contrary to price action. This is most obvious if you look at the
formation of the lower section of the price channel. Rather than indicating the
presence of an increasing bullish bias, as suggested by the conventional
indicator at the bottom of the chart, there is actually a last-minute dip in
strength that draws the channel downward. This is what gives the false
impression of increasing strength with the conventional indicator.
Remember
It’s the bias that exists between higher and lower closes that creates the price channel.
Now let’s look at point ‘B’ on the chart, starting with the conventional
indicator. In mid-September the relative channel