My system works just fine during last 5 years, including the bear market. However it doesn't produce good results when I run it during 1990-1998. Does it really matter that my backtesting doesn't work far back, but works during last 5 years and has over 500 trades? Anybody had similar experience? Since volume was different in 90th and some stocks are not even trading now then can one really judge current system on past results 10 years ago?