February 3, 2012

Lately, I've expanded my research in trying to better understand what's going on under the hood. Trying to refine and/or design more sophisticated procedures with what I simply call “interesting” properties.

As you design new procedures, you are bound to get some with what becomes desirable side effects. For instance, I like the trade clustering in my latest design (Trade Slicing) with its ability to position entries near the low of price swings not by predicting lows but as a byproduct of its scaling-in functions.

You hit some lows to a tee, not by seeking them, but as a result of sprinkling some random trades over that region of the chart. You see lows being hit as a coincidence and a consequence of the trading procedures being used.

I've presented some interesting stock charts over the past year with varying levels of performance using different trading strategies that I modified to suit my own trading methods. But what I find important in applying these enhanced payoff matrices, as in my Optimal Portfolio V note or, more recently, in On Trade Slicing, is that I get a better understanding of the ramifications of executing such trading strategies where you have exponential alpha generation.

I think Jensen was right in expressing that someone could outperform using their trading skills and investment know-how. But I also think he underestimated the concept of alpha generation. He was looking for a constant and found it. But it is not just a constant, it can be more than linear, it can be an exponential function. It is the same as saying that the Sharpe ratio is not just a ratio but that it too, becomes an exponential function concealing a Jensen alpha. The implications of achieving an exponential alpha are far-reaching and will continue to require, even on my part, more clarification and understanding.

If all that could be presented was a payoff matrix as the one in Optimal Portfolio V, one could put the whole thing on a theoretical level, a nice concept of sort, “ an if it could be done kind of thing”. And it would be justified as it would lack proof. There is only one way to show that the holding functions of the payoff matrix are more than just a theory, and that is to test them on real market data.

From my initial design in 2006 to my first paper in 2007 to my first tests in 2011, I had to find an explanation for the over-performance and elaborate the web of mathematical relationships that prevailed while the routines went on with the accumulation and trading processes. The ITRADE Formula or its siblings presented in Optimal Portfolio V or in Total Solution do describe the functions at work, but it is only through simulations over a relatively long trading interval and using a sufficient number of stocks that I could demonstrate the usefulness of my methodology.

My trading strategies are not designed to be small. However, the design is totally scalable up or down. The payoff matrix, as shown in Optimal Portfolio V and in On Trade Slicing, where a few examples are given, could, in fact, improve any trading strategy having a positive edge. It is the trading philosophy behind the payoff matrix that really matters, not necessarily which trading strategy is being used. And I think that anyone could improve his/her trading strategy by using part or all of what is described in my payoff matrix.

Since April, last year, when I started testing on real market data, each iteration, each simulation was like opening the door to more procedures that would improve on the design of the trading strategy being used. Each time trying to find the limits where the whole system would break down. And, at each iteration, finding new procedures that would enhance further desirable side effects or overall performance (see Trend or No Trend for a resumé). I still have not found where are the real limits, where do these trading methods break down? I need to know where these limits are, as I do not intend to inadvertently cross them.

None of the charts presented in my simulations could be accomplished without the underlying structure, without the mathematical foundation on which all these tests were based. I don't know what the future will bring, I don't know how you should trade, but I do know how I intend to trade.

What you see in “my” simulations is a solution to a problem that represents my point of view. My compromise to managing a multi-period and multi-asset portfolio over a long-term horizon. I leave it to each one to modify and adapt their own strategies to these concepts as I think it should help improve anyone's performance level.


Created on ... February 3, 2012,   © Guy R. Fleury. All rights reserved.