What's the point of having a system/EA that shows you (for example) 300% gain in x years, if in that period there's huge drawdowns of 50 or 70%?
Who will trade such a system? Not me, certainly.
And I think everybody would drop it, as soon as he sees his 10k account reduced to 3k.
The fact is whatever batch of "optimizing period" + "test period" (different from each other, of course) you choose... a moment arrives in which you see your account with huge drop.
I've tried:
previous 6 months optimized -> following 3 months tested;
previous 3 months optimized -> following 1 months tested;
previous 2 months optimized -> following 1 months tested;
previous 6 months optimized -> following 6 months tested;
previous 4 months optimized -> following 2 months tested.
All around the beginning of 2006. No matter what combination, at the 1st to 3rd increase in time there always huge drawdown.