Monday, September 01, 2008

The Psychology of Mechanical Trading Systems

I thought I would use the latest trading system market by Michael Bryant's Breakout Futures site to illustrate a few principles regarding the psychology of trading mechanical systems. The system is called Reliability; it's been backtested over the past ten years and traded in real time since 2004. The site makes available the full TradeStation report on Reliability. When we review the report, several features stand out. These are features that I've also found in my own recent system development:

1) Even win/loss ratio - The system generates about 50% winning trades and 50% losers. It makes money by ensuring that the average size of the winners exceeds that of the losers by about 40%. Psychologically, the trader has to accept losing on half of all occasions--even with a well-tested and conceived system.

2) Small number of trades - Over ten years, the system traded 558 times, which comes to about once a week. The system that I'm currently working on only generates a trade every few days. There just aren't that many consistent, reliable edges available in the market, which means that traders must be unusually patient in waiting for signals. The percentage of time that the system is in the market is 2.47%.

3) Max consecutive winning and losing trades - The system generated 10 consecutive winners on the long side and 6 on the short side. It also generated 8 consecutive losers on both the long and short sides. These streaks would make it easy for a trader to think he or she had a hot hand or had completely lost an edge. If traders had stopped trading after sequences of losers, they would have likely missed some of the strings of winning trades. There are many more strings of events by chance than we tend to expect.

4) Max peak-to-trough drawdown intraday - The system lost over $5000 at its peak drawdown, which is quite a hefty loss considering that the tests were constructed with trading a single ES contract. In order to benefit from the edge, the system could not set tight stops. This means that, even with an excellent edge, there tend to be significant drawdowns that are difficult to weather.

My purpose is not to suggest that the Reliability system is necessarily one that you should purchase or that it is typical of all mechanical systems (though I do respect Mike's work in general). Rather, the system embodies a few elements that illustrate how difficult it is to follow even the most sound trading rules: a large proportion of losing trades, significant time out of the market, large runs of winning and losing trades, and sizable drawdowns.

One of the arguments in favor of mechanical systems is that they eliminate emotions from trading decisions. As we can see from even a well-developed, promising system, however, it takes a fair degree of emotional resilience to stick with system rules. How much more difficult it is, then, to have the confidence to stick to setups and rules that haven't been validated!


How to F*&# Up a Trade Setup