I should have said that "it was trading". It has been all backtesting. But I haven't optimized any parameters, so I don't think that was the problem. And my first outsample worked just like the first, it wsa the second that became the problem
Not surprised, this is typical behavior of 95% of the systems I have tested. Markets change and so is the performance of mechanical systems. But without any details it's hard to discuss the specific reasons for the deterioration you observed. About 70% of the systems I have tested in the past failed forward testing due to an increase in price volatility and/or sideways moving prices. Ron
Thanks for the reply. It is a mean reversion system. I seems so weird that it survived (in theory) Through the tech bubble, the tech crash, and 9/11 before it stopped working. As someone said earlier it might just be a decrease in volitility in that time period. Oh well.
Oh well, if you've given up and the strategy is dead, could you post some of the details ... just for academic use. Maybe some of us could perform some magic and resurect some of it from the dead
I'm not quite to that point yet. I have a couple of more things that I will try before I really give up on it. Not sure if you would like it. It doesn't use any trend or momentum stuff. LOL If I do get to the point where I think it is useless, I will post it.