Developing A Futures Trading System

You’d be crazy to start trading in the futures market without first having some type of trading system in place to guide your actions. The futures market is volatile and one in which respectable amounts of money can be won or lost on just a single trade.After a bit of research, you’ll quickly realize there are a number of futures trading systems which have already been developed by “experts” and that are ready to ship right to your doorstep. But how will you know which of these trading systems will best match your trading characteristics? And more importantly, will they work at all?With a bit of effort, you can develop your own, personalized trading system; one that works perfectly for you. It’s not that difficult if you keep a few fundamentals in mind. Before building a trading system, you need to understand what you want to achieve and how to go about it. We have previously discussed trading system goals - now we will talk about actually building a system.

Step 1 - Choose a market

The market you choose depends on the amount of margin you have relative to the margin requirements, your tolerance for risk and what your broker can access. Also, remember that some overseas markets may trade during the night time - this can be an advantage or disadvantage.

Step 2 - Obtain market data

The next step is to obtain historical data for the futures market you want to trade. You will need the time or date of the data, the high, low, close and open price, the volume and the open interest. This data can be obtained from price data vendors, from your broker, or even from the exchange. Unless you like rekeying information, get it in electronic format.When a contract nears expiry, the open interest decreases and the next contract becomes more active. You need to use the data for the most heavily traded contract, whilst ensuring no sudden jumps or drops in prices as you move from one contract to another.You’ll need as much historical data as possible, with a minimum of 2 years worth for end of day data.

Step 3 - Develop a trading model

All trading models are based on the trader’s view of what defines entry and exit points. Here are some common approaches:

• Breakout - the market falls below, or rises above the low or high for a certain number of periods
• Momentum change - the rate of increase or decrease in prices change
• Smoothing - a moving average of the data turns up or down, or a long and short moving average cross
• Pattern recognition - the chart shows a recurring pattern such as “head and shoulders” or a “pennant”, or by using genetic algorithms to recognise price patterns
• Prediction - attempting to mathematically forecast the next day’s price levels, for example using neural networks

 The common factor is that all these approaches try to distinguish meaningful price action from random market noise. You need to choose one that you are comfortable with, that suits the way you think.My suggestion is to use either a breakout or a smoothing approach. Both generate trading signals by isolating trend changes from random price movements.

Step 4 - Backtest your system

This is a matter of generating trading signals against your test data, and then seeing how they work out. Tools like TradeStation can help you test your system, but you could also use Excel.Remember to take slippage and brokerage commissions into account. Also, you should test your system against different data from that which you used to develop it. Also, test it across data that shows a variety of market conditions, preferably over a number of years including both up and down markets.Make sure you don’t make silly errors such as buying at the day’s low and selling at the high. You’ll need an approach that will work in real life. If in doubt, get someone else to check it out for you. A good system should work with different markets (although not optimally).You need to evaluate the expectancy (expected win per trade), the maximum equity drawdown, the number of consecutive wins and losses and the number of trades. Avoid a system that trades too often, or has a lot of ups and downs.

Step 5 - Adjust your system

If your trading system doesn’t quite perform as you would have hoped, you may be able to tweak it. However, you should not look at each trade that it failed in, then concoct a new rule to accomodate that situation. The simpler a model is, the more robust it will be. Simple means as few rules as possible.Similarly, be careful of curve fitting. This is tweaking all the parameters so that it gives an optimal result on the test set. If you do this, it is likely to not work so well in real life. That is why I said to test your system against data that it was not developed with. Keep your test set and development sets of data separate.

Start trading

You’ve thoroughly tested your system and you are comfortable with the expected results. You understand how it works. Now you need to start trading.The important thing is to not be discouraged if you have some losses. You need to be confident with your system provided you have developed and tested it properly.Good luck with your trading system development!