September 8, 2012
I occasionally participate in the Automated Trading Strategies forum on LinkedIn. And over the past few weeks, I provided some comments which elaborate on the trading methods I use in my strategy design. The following observations are almost in chronological order.
Everyone has a trading method. However, whatever it is, it must do the job, it must deliver. Otherwise, why spend so much time and effort to produce under-performers.
Posted Aug. 26, 2012
In previous posts, I expressed that the main reason some trading strategies fail going forward is simply due to poor or bad strategy designs. To illustrate the point, I present the worst script I could find which was published in Active Trader magazine in November 2006. At the time the SixSignal script was touted as a “wow” script, having great performance over past data.
(click to enlarge)
Here are the current numbers: it has an APR of - 64.56% per year; an 84.90% exposure rate and a dismal 99.69% drawdown. The SixSignal script is not that complicated; it is a variation on a theme and was doomed from the start, but nonetheless was published for novice consumption. Anyone adopting that particular script over that last 6 years would have seen a $100,000 portfolio meltdown to $310. I don't think anyone could recuperate from there.
One should study what is available (since a lot is already free out there). At least it can tell you hundreds of ways on how not to do it. There are thousands of trading methods that have been tested over the past 50 years, you don't have to redesign everything from scratch. For sure, you can save a lot of time by looking over somebody else's shoulder; testing their designs and at least maybe save yourself a lot of misery.
The next day one expressed the notion that if the strategy was so bad then just by reversing the trading logic would make it a profitable proposition. To which I replied:
It is an interesting observation. At first view, I certainly would agree. I often look for this kind of trading methods in the hope of reversing them to profit. However, this particular script is part of the exceptions: I think it would do as bad even when reversed. I would have to test it to be sure, but my comprehension of this script in its present state leads me to that conclusion. However, I do think that changing part of its trading methodology could reverse its losing ways.
After a few hours work to reversed the SixSignal trading logic, I posted the following (Aug. 27th):
I reversed the logic of the SixSignal November 2006 script. As expected, it did not improve performance, on the contrary, it got worst:
(click to enlarge)
It might seem surprising at first, but when analyzing the trading methods used, the script was shooting itself in the foot, and as if it wasn't enough, with a double barrel.
I'll do some work on the script to change its trading methods just in an effort to show that even badly designed scripts can improve if you can liberate them from flaws and misconceptions. So give me a couple of hours and I'll be back.
That was a tall order. The SixSignal script was published in a trade magazine as a best of the breed, as a “you can make some money using this script” kind of thing. I would argue that the author or the publisher did not properly test the script before publishing. I certainly find it a disservice to publish such a misconceived script. IMHO, the author of a trading script is somehow responsible for the consequences to others; since using their published script might literally destroy someone's portfolio. So, my recommendation, whatever published script you see, test it exhaustively before use. As a matter of fact, do this for all the scripts you may design.
Again posted Aug. 27, 2012
The chart below shows my modifications to the SixSignal script. I can do better, but 2 hours... is not that long.
The first thing I did was to stop the SixSignal script from selling short; the script is not really good at it. And then I started to improve on how it perceived its positions in the market.
(click to enlarge)
What should be noted from the above chart is that it did become profitable. It traded a little bit more over its 6 years simulation; and converted a losing proposition into a system that is trying to accumulate shares as well as trade over the accumulation process which is what I advocate as a method to increase long-term performance.
Of note in the performance metrics: you have 87% of trades showing a profit. And the average profit per trade is 10.9 times average lost. This, to me, is a relatively fast action script having an average holding period for losing positions of only 10 days whereas longs had on average a more reasonable holding interval of 219.85 trading days. I usually tend to hold on longer for hopefully more profits.
The above chart is all I see of the data (about 220 trading days out of 1,500). I design trading procedures from a mental picture of what they should do since I do not have access to the whole trading history. I am forced to visualize what my trading procedures would do over unpredictable price fluctuations. I find it an advantage, not a handicap since it forces me to design scripts that do not necessarily know what is coming their way.
What I advocate can be said in a single sentence: trade over your cumulative Buy & Hold process. Use the trading profits and the increasing equity to feed the stock accumulation program.
For me, all this represents a small step, a slight modification to an existing equation. Since I can represent all the generated profits from any trading strategy using an enhanced payoff matrix: Σ(H+.*ΔP). The improvement will translate into Σ(H(1+g)(t-1).*ΔP); which will result in exponential profit growth.
What this note emphasizes is that you can take a bad script like SixSignal and convert it to have an accumulative stance over time; recycle trading profits to trade more and accumulate more shares; from this process, IMHO, you should see your own scripts transformed in this manner.
It all translates into a series of trading procedures; a series of what to do when... such and such occurs. Of the three scripts used in this note: the original SixSignal, its reversed or the improved version; they are just a series of program instructions to be executed in the order given. Nothing more. It is the trading strategy being executed that makes the difference in performance.
Created... September 08, 2012, © Guy R. Fleury. All rights reserved