This is a post I wrote for one of the Forex forums few years ago. It was under a screen name, but it is mine. No editing was done so it has a little different feel that my current work. Subject was backtesting and the thread was full of opinions that it doesn’t work, meaning it is a waste of time.  In the future I will expand on this topic and cover it more in depth.

“Hopefully if one decides to trade one believes that he-she has an edge. There is no other way to prove or disprove it (other then start trading) but to go through historical data and see if that worked in the past – backtesting . When you get an idea on trading method do you just go ahead and start trading it? There is however a lot about backtesting that most people do not understand. Some ideas are simply not suitable to backtesting.

Anything like divergences are simply too subjective and ambiguous to be backtested either by computer or by hand. But plenty of  ideas can be easily  backtested as far back as you want. To illustrate the point let’s say that on daily charts we will go with direction of 50 SMA. If the daily bar closes above MA buy opening of the next bar and sell(cover) close. Do it for every day until bar closes under 50 SMA when you simply sell open and buy(cover) close. I’m not saying this particular idea works only that it can be successfully backtested.

Another thing is you have to have data to backtest far enough to make the outcome a viable sample. I wouldn’t bother backtesting anything if I can’t see an outcome of at least 500 CONSECUTIVE trades, and preferably much more(1000-2000 trades). Hopefully that time span would cover all kinds of market conditions from raging, running markets to absolute non-movement and anything in between.

If your idea showed some promise for given time sample than there are some other things to consider. Were the loosing periods survivable in terms of time and money drawdowns? What good is a methodology if you need a 100,000$ to trade one lot and has two year span of losses?
Backtesting works for me, as I believe that market behaviours will repeat themselves, not in mirror image perhaps but close enough to work out an edge.

By the way, why would you consider trading any kind of approach if you didn’t have some kind of precedent from past that you can make money that way? If you were to trade fundamental announcements,for example, you would only do it because you “know” that markets respond to them and you “know” it only because it happened in the past. As a matter of fact any trading methodology is based on some previuos experiences and precedences- or in a nutshell backtesting. “

Mike K.