June 18, 2014

My next trading strategy to be analyzed is kind of another strange trading script. It buys and sells on about every price swing. It sets a no-trade zone. Will buy below and sell above. Yet all entries are the result of random functions. It gives the illusion of perfect timing when, in fact, trades are coincidental. Meaning that, on purpose, they are not hitting the highs and the lows but are a side effect and a direct consequence of the methodology used.

It's the trading mechanics, the trading procedures themselves, that make this trading script prosper. After all, a program is just a program, a series of command statements to be executed by your computer, even if set to operate over some 25 years. It took me a few days just to find this script out of the 200+ programs on my machine.

In general, this trading script behaves something like this:

General Trading Behavior (DEVX V3)

Swinging It

(click to enlarge)

What's remarkable here are the entries and exits (blue and red arrows). They are relatively well distributed as a result of their scaling in and out functions.

This program operates on the premise that position entry prices do not matter that much since all its entries are the result of random functions. This implies no indicator of any kind is being used to determine entry prices except the outcome of a random function, naturally within the rules of engagement.

Rules of engagement were designed, and the rest is left to the outcome of the random functions and to which path each stock's price action would take. It's quite a sophisticated and intricate program (950 lines of Wealth-Lab code which is a lot considering).

This strategy could also be improved to perform even better. My present dilemma is: do I improve it now, or do I use this 2½-year-old trading script as is? The modifications could take some time. It's a complex program. But on the other hand, doing the modifications could bring it to another level altogether. Just incorporating what was learned in the last two 25-year tests could be enough in itself to gain the number one spot on my most preferred list.

Both previous strategies were tested on their never-before-seen 30 DOW stocks over the past 25 years. From these two tests, I already know what this 3rd program is going to do, even before performing this new series of tests. My real question is: will it reach the top of the list or arrive in second place?

I really like this script for the way it behaves over time. Just a look at the snapshot above, and one could easily surmise that it will be generating profits, albeit in a different manner than the other two, but isn't that why we design different trading strategies in the first place? Isn't it to reach the same objectives or better but take different paths? And then comparing which we like best based, on our personal tastes and propensities.

These 3 trading scripts use different trading methods to accomplish their respective long-term goal, which is to accumulate shares for the long term and trade over the process. Only Mod01 of the BBB System was coerced to liquidate its inventory from time to time. The other two just want to accumulate shares for the long term and trade all they can over the process.

Trading Behavior Compared

The general trading behavior can be easily compared by showing a sub-series of trading action over even a relatively short period of time:





General Entry and Exit Behavior

LMK 0.06


May-June 2011

25 years

30 stocks

LMK 0.06

BBB Mod 01 


May 2011

25 years

30 stocks

BBB Mod 01



(Nov. 2011 -

Jan. 2012)

6 years

11 stocks


Each of these 3 trading strategies was designed to behave differently, yet each could achieve interesting long-term performance results. The one still to be tested over the last 25 years is DEVX V3.

The two previous tests were done under the same testing conditions: same initial capital, same stocks (meaning same price series), and same time interval. Therefore this third test will have to be done under the same conditions. This way, it will be easy to compare all three. Only IBM would have been seen by this modified trading script (and indirectly, by one of its variations) for a 6-year period ending in early 2012.

One of the features of this third program is its "mood" controlling vector, which could be extracted from the program to provide some outside control over its future objectives. You could say things like starting tomorrow: increase/decrease general trading activity, overweight stocks 3, 7, and 12, increase inventory distribution on 19, push for more profits on highest performers, underweight stocks 11, 18, and 22, or simply: steady as it goes. All from outside the program by providing the appropriate input to the control vector. It would translate the changes in trading policies and objectives and would adapt to them going forward, giving human control over policy orientation and future portfolio performance. Imagine the avenues of research it could open.

Wide Open Doors

By opening this door, you also increase the number of different kinds of tests to be made. All dealing with different trading procedures having long-term perspectives.

Some might think that what I propose or the tests I do are just BS. I can easily understand that. I would have the same first reaction to things that go beyond what I can do or that I might not be interested in doing. What I provide is not a "black box", it's a framework, principles on which anyone, on their own, could design profitable trading strategies going forward. There are so many ways to design worthwhile trading strategies.

My question is: How could I apply a trading strategy with any kind of conviction if I don't know what it is really doing and why? At least, if I do a backtest on the trading principles involved, I can study the intricacies and footprints left behind by all those trading procedures. I can see what would have happened and then put my bean-counting abilities to work in order to provide a better understanding of the whole process.

I see it all as so simple, a use of common sense and the acceptance of some compromises. I know I can't get it all at the same time. I know that randomness plays a major role in stock price movements. I know I can't predict the future. I know that about anyone can do better than I can. I also know that portfolio management is a long-term endeavor. I even almost know that I should not be able to design such trading strategies; yet when I test these trading concepts over extended periods of time (20+ years), I get more than just interesting results. I view these as a simple consequence of the applied trading methodology. Something kind of a default mode. You win because you designed your trading strategy to win from the beginning.

Therefore, for a conclusion, as well as for the need for comparisons, I'll have to do several tests. The first of which will have to be based on the original trading script published in 2002 on which this is based, followed by my selected modified version of this trading script with bet sizes of $100, $1,000, and $5,000 in order to compare with previous tests. The selected modified version is one of five I've found on the same theme. I've elected to test and then improve on the first one found. These 5 programs were designed from November 2011 to January 2012. They are old trading scripts, part of my continued search to do better. Which one is best? I don't know. I would have to test them all and then make a choice. That is quite a lot of work, and being alone to do it all will take some time. Since I intend to improve performance on the selected script, it might not matter that much which of the 5 versions is used for testing purposes.

This is not the first time that this trading strategy, or one of its variations, has been discussed or illustrated in my research notes. For instance, a variation on this script's theme was highlighted in my article on trade slicing. Another variation, or maybe the same, was used in my January 2012 presentation on my trading methodology. You can also find part of the explanation of why this kind of random entry trading system works in the following research note.

Therefore, all this is not new. It's only that this chosen trading strategy, a variation on a theme, will now be tested over a stock selection it has not seen over a testing period of 25 years which also goes way beyond its testing ground. During the development of these 5 trading strategies, only 11 stocks had been tested over the previous 6 years of data, where only one stock (IBM) would have seen 1 of the 5 strategies. So what is coming is, in itself, quite a challenge for an old trading script.

Created... June 18, 2014,    © Guy R. Fleury. All rights reserved.