- 1998: Volume 7, No. 5 Market Timing Key to System Performance

Many fund managers and professional traders believe that computerized systems that work consistently in real-time can only be expected to perform at 40 percent or less winning trades. They believe that searching for the Holy Grail or anything close to it can only be an exercise in curve fitting (i.e. fitting trades to past data behavior using many optimized parameters) and is, therefore, futile. The fact is that claims of very high percentage winning trades in advertisements from system development companies are often found to be misleading, for reasons such as, lack of adequate back-testing, curve-fitting or continued over-optimization.

These generally held beliefs are a hindrance to the system development community, as one is led to believe that trying to produce a system with a very low drawdown, a high percentage of winning trades and a good profit factor is impossible, except on past data. Many computerized systems are also not accessible to smaller traders as the account size required to trade the system is often too large, due to the predicted drawdown and the necessity of trading many products. Moreover, the timescale of the trades is often very long and psychologically the system becomes very taxing to trade.

Trading Systems: The Catch

The bulk of these 40 percent or so winning trade systems are, in some form or another, trend-following. It is a widely-based belief that trend-following is the main—if not the only—way to trade professionally.

One of the problems with trend-following is that many markets can simultaneously go into fairly static periods for months at a time. This can cause fairly horrendous drawdowns, caused by the generation of false signals, leaving the faint-hearted or lesser-financed in serious jeopardy.

Many traders abandon their systems after three or more losing trades. In some cases, they are right to do so but, sometimes, this is just before the markets begin to trend again and the system resumes its performance. I say that, in some cases, they are right to do so because there can never be any guarantee that the system will return to its previous back-tested performance.

So where does that leave the system trader? Perhaps an answer may be to historically test how long these non-trending periods can last and if they can seem to be at relatively regular intervals. From much analysis, I conclude that certain markets can be timed to predict when there is a high probability of the market tredning.

From this analysis, one could, in theory, switch between a trend-following and a contrarian system, all within a single computerized signal-generating system. I have found that using this technique can considerably improve the trend-following system's winning trade percentage, as it will only become active when the market has a high likelihood of beginning to trend. Of course, this reduces false signals greatly.

Using the same historically tested timings for trend occurrence, if a trend is not yet due and the previous trend has faded, is the market very likely to range (i.e. create a wave form)? If this is the case, would not a reliable measure of decreasing momentum be able to indicate a wave top—and therefore a sell signal (and vice versa for a buy signal)? The answer is that there is a high degree of probability.

The Importance of Data

There are many pitfalls when one back-tests ideas using computers. One needs a continuous stream of data going back many years. Individual contract months are not suitable, even continuous nearest futures have serious problems and cannot reflect true trading. It is advisable to trade the most liquid contract. Certain products on the market, such as the CRB InfoTech CD with CRB end-of-day data enable a system developer and trader to use multiple forms of continuous contracts, enabling back-adjustment to reflect results as if they were really traded.

Trading With the Right Broker

It is not only for system development where the right format of data is required, but also to trade a system to produce results consistent with back-testing. Consistency is the key. I have found that close contact with, and a full understanding of the system by, the right broker are invaluable during the development phase and even more so while trading the system. This means that you get the fill and exit that the system needs in order to perform correctly. This is harder than it might sound. Finding the right broker is difficult. My own system can be traded through my broker at Angus Jackson Inc. This ensures a high degree of consistency.

Based on these, and other, principles, I have been able to create systems that have relatively low drawdowns, a high degree of profitability and —psychologically important—a high percentage of winning trades. However, crucially, they are also working in real-time and making profits.

Indicators

A reliable way of testing and indicator is to turn it into a back-testable system. Most popular indicators when turned into a system do not work and, if traded consistently on their own, could easily lose you money hand over fist. If most indicators are at all reliable, why are they not routinely sold as back-testable systems? The indicators I use come only from formulae within my systems that are proven to work. I use them merely to give me an early warning of coming signals and a deeper understanding of the system I'm trading. For those who enjoy making their own trading decisions, I suggest they consider system-tested indicators, paint bars and showmes, available from myself, as an addition to their trading methodology.

Select Trading System Benefits

• Low drawdown ideal for small traders.
• Trade with as little as a $5,000 account.
• Quick confirmation of systems' abilities.
• Let top professionals execute your trades.
• Stressful wrong decisions? Relax—they're gone.
• Indecision in pulling the trigger? No problem.
• Hire systems for $300 or less per quarter.
• Alternatively, indicators proven by back-testing.
• Long-term commitment to keep the system? Forget it.

Derek Howard is a trader and system developer and proprietor of Select Trading. He can be contacted by e-mail at Select1@netcomuk.co.uk or by fax on: Int. +44 (0) 1303 814491. U.S. inquiries may contact Michael E. Rose at Angus Jackson Inc by phone at 954-772-1166 or by e-mail at mike@angusjackson.com