Analyzing Stress Testing Results

[tradingGary]

I have been building a strategy where one of the conditions is a crossover based on fast and slow moving averages and finally have decent results with a testing period over the last 5 years. My question to anyone that has experience with these type of strategies is:
When I do stress testing and change the time period of either fast/slow time period by roughly 10 percent, I get poor results where the profit and sharpe ratio are significantly lower (50-75% lower).

Is this a typical of cross-over strategy behavior, or does this imply that I have curve fitted my period? (I did use in-sample and out of sample to optimize this strategy)

Any input on this is greatly appreciated.
Thank you!
Gary

[NinjaTrader_Bertrand]

Gary, it's hard to tell conclusively - you would need to monitor going forward : it could be really determined the optimal period to use to catch the dominant cyclic aspects of this market, however crossover strategies typically are very sensitive to settings used and you would need to consider that after a lot of testing the out of sample data becomes easily in sample data again as you're 'fitting' the parameters to hold up on the out of sample data...better still than not using out of sample testing...agreed here.

[tradingGary]

Thanks Bertrand.

Does anyone else have similar experience with CrossOver SMA time period settings?