Conducting and Evaluating a Forward Test

[selfmade]

I've developed a trend-following trading system for the futures market using Tradestation's EasyLanguage. Per the backtest, the system will open on average between 5 and 15 trades a year for each market traded. The system was tested on over 20 years of historical data using a daily chart, across a wide variety of markets, and the backtest results look promising.

I am starting a forward test on this system using a demo account and live market data. Like most trend-following systems, this one can suffer sizable drawdowns. For a system like this, I think it is possible to have a drawdown right off the bat in a forward test, but still have an overall good algorithm. My question is how do I interpret the forward test results (distinguishing between drawdowns and system failure), and for how long should I forward test the system before going live with real capital?

 

[Rhody Trader]

The problem with a trend following system that only kicks out 5-15 trades per year is that it's going to take you a long time to collect forward tested results sufficient to draw any conclusion. Here's the good news, though. The whole idea of a trend trading system is to identify prospective new trends and ride them for all they're worth. It's pretty easy to see whether the system is doing that or not. All you have to do is look at the trends that have taken place and see whether you system caught them or not.

 

[selfmade]

So if the system visually catches trends in the forward test like it did in the backtest, then I should move forward with live trading?

 

[Rhody Trader]

You don't have any control over what the market does, of course, but if the system is doing what it's supposed to be doing then you might as well.

 

[selfmade]

Good response. Thank you!