Ask Colin

Is there a general approach to testing a trading plan?

The detailed question was:

I have finally put together a basic trading plan. I now need to test this plan but am unsure how to adequately do this. I haven’t come across any books that address how one should thoroughly test a trading plan. I don’t have any software I can use to back test so am assuming I need to manually select all my test cases. How do I go about selecting specific cases to test? Is there a general approach to testing a trading plan?

This is an interesting question. It is one that I have not given a huge amount of thought.

There is charting software that allows you to back-test a trading method. It is ideally used where you have what would usually be described as a system. Such a system would have clear rules that you can define mathematically. However, if you have a method that leaves room for discretion, then software is not going to help you as far as system testing is concerned.

 I developed and tested my methods was in the days before computerised charting when we drew charts by hand. I have described elsewhere how I did it, but essentially I took a stock I had never charted before and drew its chart from years of back copies of newspapers day by day. I marked my decisions if any on the chart before I turned over the next day of newspaper data. This is as close to paper trading as you can get without waiting 24 hours between each plot.

 Today I would use charting software to do this, because you can now do the same thing in Insight Trader charting software – put a day of data on the screen, then add more data a day at a time, marking decisions as you go. This saves an enormous amount of drudgery drawing charts by hand and is just as effective.

If you do not have any software at all, I suggest that you are going to have to get some historical data and do what I used to do. In addition, you should paper trade some stocks going forward in real time.

The general ideas of how you go about this is to select for testing a wide range of the kinds of stocks that you intend investing in or trading. Then, the longer the time period you can test over the better. In particular, you want to make sure that you have both bull and bear market experience in the testing. The more stocks you can test and the more randomly you select the stocks to test over, the better.

These are really just some general ideas. I am a long way from being an expert in this area. A book that I would recommend on testing systems is Le Beau and Lucas Computer Analysis of the Futures Market. Don’t worry that the book says futures and you are a stock investor/trader. The principles are the same for any market.