6-year backtest looks good, and even better with timefilter - page 2

 
If you're still looking for Kelly and/or Optimal-f calculations, review my codes https://www.mql5.com/en/forum/137889. My opinion on this thread, if you think your system can with-stand the future and you don't believe it to be curve fitted, then try running it upon couple of different currency pairs for about 6 years period. If you think my suggestion is a really bad idea because your system is optimized for Eurusd then perhaps you should ask yourself why it's not curve-fitted.
 

Its probably curve fitted. As soon as I get time I'll adjust it on 2 years, then run it on 2-3 different 2-year segments of history.

I did something similar a while back, when I ran it over 6 years the performance was so-so... but in that 2-year segment in the middle of

the test the curve shot straight up

I have a gut feeling that whatever we do and no matter how good the test looks, the (any) EA is going to fail miserably when trading forward.

Reason: