I think that’s true. That’s why I’m willing to accept systems with somewhat lower theoretical performance if I think they have the property of being different from what I believe most other system traders are using. When I raise the point with would-be system designers that much historical research may be invalidated by the changing nature of futures markets, they invariably reply that the solution is to develop systems based on recent data—as if it were that easy. There’s a serious problem with this approach. Recent data has to be less statistically significant than long-term historical data
...more
