Neil Weintraub

Trading Chicago Style

(Insights and Strategies of Today's Top Traders)

MICHAEL CHALEK

Software Without Tears

Mr. Chalek attended the University of Illinois and received a B.S. degree in electrical engineering. In July of 1979 he received a graduate M.S. in electrical engineering. Mr. Chalek has also done additional graduate studies in microprocessor theory and design, leading toward a doctorate in electrical engineering.

Mike: I began trading the stock market shortly after graduating from the University of Illinois (BSEE) in 1973. At the time I did not know that what a bear market was or that we were in one. My unskilled experience led to some very gut wrenching experiences with several stocks. I knew nothing of "selling short", just buying stocks. I accumulated shares in Teleprompter, Combustion Engineering, Pizza Hut, Williams Company, and a few others along with an inept stockbroker. Within 6 months I lost my entire savings of $5,000. It was time to hit the books. My first encounter was Nicholas Darvas' How I Made $2,000,000 in the Stock Market. It was actually the best book to start with. Both sides of the trading equation (technical and fundamental) were touched on. With my engineering background, it didn't take me long to go with the underlying principle of technical analysis for possible trading success. I moved on to other technical books such as Technical Analysis of Stocks and Commodities and Joe Granville's A Strategy of Daily Stock Market Timing. During the winter of 1977 I went to the Chicago Stacks (exchange libraries) after work, and painstakingly, by hand, I proved Granville's theory of On Balance Volume using all of the Dow Jones Stocks in the 1929 Wall Street Journal.

During 1983 I received a phone call from an individual (Allen Ko) who had read an article about me and was interested in some of my work. Over the course of about two years, Ko and I became good friends, and soon we codeveloped a system involving an adaptable volatility variable and a unique price pattern. Between the years of 1984 and 1987, Ko parlayed $25,000 to over $2,000,000. I probably only made about $500,000 during that time. My discipline in trading was not as good as his. Sticking to a system was extremely hard, especially during the drawdown periods.

Mike: Dual Thrust is a function of dynamic volatility and a unique price structured pattern which, when used collectively, gives inherent support and resistance levels in the market. We are looking for a breakout with either of these points and treat the signals as a stop and reversal methodology. The system appears to work rather well on the S&P intraday charts using time periods of 30 minutes or more.

Mike: A pit trader could possibly use it if he had a computer on the side, say one to two minute bars. This would only be good for a fast scalp. Not much could be said for a fundamental trader because this is mathematical methodology.

Mike: Choosing software to trade is a very tricky task. Do not rely solely on such third-party reviews as Futures Truth. They are only part of the equation. The user should ask the software provider what the data source when the software model was backtested. This might also reveal such things as whether the developer used actual contract data as opposed to continuous generated contract data. I have found vast differences when applying the two approaches. My software has a rollover feature, which painstakingly applies actual contract data and will ask the user when to rollover into the expiration month. TradeStation does not provide that kind of option. Watch out! The aspect of drawdown can mean different things to different people. I have found that some software developers actually use closed position drawdown and not mark to market open drawdown, which simulates a real life situation. A good software package will provide an ASCII data option, since most data vendors allow you to convert their particular format into ASCII data. This is a universally accepted data format, which can be readily imported into any text editor.

Mike: Developers should provide full disclosure of their methodologies. Avoid those "black boxes". I have found that most vendors generate their code into a black box when they are either copying a strategy from a book or curve-fitting a specific trade occurrence somewhere in the past. By generating a "white-box", the end-user will experience a higher comfort level. Otherwise, you simply can't trust the software. In addition, if the programmer had errors in the software, you would have no way to cross check his results. It is important to know how many parameters the system is using. Anyone can take a methodology and curve fit it to his or her liking by using an over abundance of variables to address a number of market conditions. These conditions will change in the future; they often do. I have found over the years that if a system has more than five variables, it falls apart soon after its release. Just peruse the Future Truth rankings and see for yourself. Dual Thrust uses four variables and I consider that a limit. As an extension to this argument, ask the developer to provide a three-dimensional profit curve for a wide range selection of his variables. This is a good way to visually see if you have a good robust model or just some isolated area of profit surrounded by an abyss of losses. In the February/March 1998 issue of Futures Truth, Dual Thrust was put to the real test. They applied it to 60-plus domestic and foreign markets. On top of that, they chose only one parameter set for all of the markets and found it to be a very robust system by performing in almost all of the markets.

The bottom-line real test would be for the purchaser of the software to ask the developer for any real-time results of anybody using the model. That should flush out about 90% of the noise factor. Unfortunately, with the advent of TradeStation, there has been a proliferation of software developers; the number has risen exponentially over the years. TradeStation has several bugs in it, which have not been resolved to date. There is some Monday morning quarterbacking going on in that some unscrupulous developers have been taking advantage of this weakness. Supposedly this is being taken care of in Version 5.0. Therefore, the platform for developing the software model is very important to consider.

Another facet to the software model would be the amount of markets that it can trade. There are too many systems out there that only trade one market. A more robust system will trade several markets, although that could be considered by some to be a form of over-optimization.