That's why the post is called "more food" - for thought. imo, I would just guess probably nearly every trader, knowingly or unknowingly, has done some sort(s) of backtesting (with whatever approaches), or its equivalence if not more, whether correctly/ fully or not.
Hey: I am on vacation for a couple of days, but these posts caught my attention. It's encouraging to see that Oddtrader thought enough about the subject to look up Acrary's posts. He has posted reference to the one that really makes my case as far as how research should be done. Now all you have to do is take the next step. Just think for a moment about what is implied in Acrary's comments. You can backtest, find that the "system" you tried looks good, tests good, and then is a loser realtime, or you can begin by testing variables for corellation and dependency, and THEN backtesting. Sure its possible that you still end up with a net loser. In fact if you do it right (Monte Carlo test), one of the pieces of info that you end up with is an estimate of how likely it is that your system is a loser. You guys have a nice holiday. Steve46 P.S. Regarding Mind's idea, using a Monte Carlo engine could speed up the test process quite a bit ("@Risk" or equivalent software). You could probably determine whether that set of data was representative in about 30 minutes, rather than papertrading it for months. Just a thought.