@ironchef, what you could also do is to cut your dataset in eg 4 parts, and test how your optimized/curvefitted stoploss works out.
That would make of a simplified version of k-fold cross validation, simpler, but better then not doing it. And pragmatic, for us, retail
Btw since we seem to trade a bit in the same way, swing / portfolio managing using options, it should be good to exchange ideas, without giving away your hard work ofcourse. I do something similar perhaps, basically it’s derived from Muppets advice (theta/gamma)