Computers are among the most common pieces of equipment that traders use. Unfortunately, the vast majority of traders are engaging their computer to enter into trades, without engaging their mind to think! There is a huge difference between buying a computer trading system and sitting down to do the research to develop your own computer-based trading system. Any computer-based trading system that you buy might be profitable (in sales or trading) to its author; however, to a trader that is not intimately familiar with the research behind it, and why the rules exist, the program is basically worthless. The vast majority of successful traders developed a methodology by building their own trading system from the ground up. As they devise their trading strategy, they very well could have used indicators, beliefs, and subsystem rules by purchasing them, studying them, or learning them from a more experienced trader. The distinction is that until the techniques have been verified, studied, and invariably modified, the trader will not use them. Only after constant experimentation with the new idea, and after verifying and internalizing, and then deciding that the idea is valid and valuable, does the trader add the new concept to his or her methodology.
Unfortunately, designing a computer trading system that accurately reflects your trading methodology demands a lot of time. It is not something that you can put together over a weekend. However, the huge advantage is that when you are done, you have accomplished something that 98 percent of all traders never do. Consequently you will see your trades produce consistent results. Once you are producing consistent results, either profitable or not, you can begin working on your ability to perceive the market better. As your perception increases, your ability to produce consistent profits will increase. It should be stressed that in order to have the internal beliefs required to do the research necessary to develop your methodology, you must make a decision to become responsible for all your beliefs. At some point all traders must confront how they will use the power of the computer to optimize their trading methodology. Optimization is achieved when you have written a mathematical formula or theory that describes the market action (in part or wholly) through variables. By programming the computer to literally perform all the mathematical permutations possible on the variables, and then correlating these permutations to the profitability, you can determine the variables that created the most profit. In other words, by determining the best combination of variables to maximize profitability, you can create a highly profitable methodology. The only problem is that it is good only for historical data; it is absolutely worthless in real time.
Say we have a simple moving-average crossover trading system. Our rules are very simple. First, if the short-period moving average goes above the longer-term moving average, go long. Second, if the shorter moving average goes under the longer moving average, go short. Consequently we are always long or short. By writing (or buying) a program, we can specify that we want to vary the shorter moving average from a period of 2 to 20, and the longer period from 21 to 60. Then by allowing the computer to test all the permutations that could occur by varying the periods of the shorter and the longer moving average, and by keeping track of the profitability, we can determine the most profitable short-term and long-term moving average. For example, the computer might indicate that a shorter moving average of 15 days and a longer-term moving average of 57 days generates the best profit. Typically the second most profitable combination of variables will generate less than half as much profits as the most profitable combination! At this point most beginning traders are very excited, convinced that they have just found the Holy Grail! There is a huge problem here. These traders have just wasted some very valuable time programming, because all they have accomplished is to curve-fit their variables to historical data. While it appears to be an outstanding combination of variables, it is an outstanding combination because it is only looking at the specific data used to perform the permutations. In other words, if they modified the data by changing either the dates used or the contracts, and re-performed the computer optimization study, they would come up with different short-term and long-term moving average values.
All traders use optimization studies to one degree or another. It is important to realize that by varying the length of the data used and by using different contracts, the value of your variables will vary. If you do in fact optimize your indicators and trading system, your goal is to find a group of variables that perform equally well on different contracts and different time periods. When you start analyzing the profitability of the various variables, you should automatically discard the variables that generated profits far in excess of any other profitable variables. Why? Suppose that a certain set of variables generated profits of $2000 and the second most profitable set of variables generated profits of $1000, and the third most profitable set of variables generated profits of $950. Then it would be safe to say that the variables that generated the profits of $2000 are so optimized that they are worthless. Your goal whenever you are doing optimization studies is to come up with a set of variables that perform equally well on different commodities, using a wide variety of different data lengths and, perhaps most important, using commodities that are clearly in bull and bear markets. Lately there have been some very good computer-based systems that have generated profits. Typically, however, the system is geared only toward a bull market. When the market goes sideways or actually drops, the system loses a lot of money. It is important as you devise your system to look at the widest possible variety of markets, trends, and time frames. When you are developing your methodology, keep in mind that you will be trading in markets dominated by bulls, markets dominated by bears, and markets where everyone is snoozing. You want your trading methodology to reflect this fact. An outstanding methodology will be profitable in all markets, in all time frames, and across all trends.