Automated futures trading, at its most sophisticated, is done entirely by one or more computers. The computer monitors the market being traded, determines what size to trade, and decides when to enter the market and when to exit. An easier-to-achieve version is to have the computer tell the trader when to enter and exit, with the trader entering the orders and determining the size to trade.
Thorough testing of the automated futures trading software is of the utmost importance. Backtesting must be done using both bootstrap and Monte Carlo techniques. The software should then run on real-time data, but without the ability to place the trade. One of the many things to check for is whether it is possible to enter or exit the market at the price the software "thinks" it has entered. Some commercial software shows entries and exits where no trades took place.
It is best to code trading size into the program. Sometimes flawed sizing code will be revealed when a system that tests profitable for one contract goes broke using the programmed sizing. A thorough testing may also reveal losses that are larger than expected along the way to final profitability. Loss along the way to final profitability is referred to as "drawdown, and is important both financially and psychologically. Unexpectedly large drawdowns have caused many a trader to abandon his system just before being rescued by a string of profitable trades.
When both the entry and exit portion and the sizing portion of the automated futures trading software have been diligently tested and look good, the trader should evaluate the software’s trading edge. The data the software needs to generate are: the percentage of wins (%W), the average win (AvgW), the average loss (AvgL) and the largest loss. The trading edge equals %W*AvgW – (1-%W)*AvgL. The term of art for that equation is “mathematical expectation.”
If at this point the system is profitable, the trading edge will be positive; however, the size of the edge is important. If the automated futures trading system has an edge of only $1 or $2 per trade, many contracts will need to be traded every day to generate satisfactory results. Even then, if something goes wrong, the system could give back all its winnings in one unexpected loss. A trading edge of $10 per trade, net of commissions and costs, is near the minimum a trader can afford to accept.