Hello Dun,

If there is a performance gain, it must be from something other.

I'm currently finalizing a new version, which will come with a more advanced Walk Forward and 7% faster backtester, but it is not yet uploaded.

Thank you.

OK, a theory: it did not occur to me before, partly as I don't know enough about the programming to have any insight, but about the only change I can think of which might be significant enough to cause that much of a difference in performance, is that I have greatly reduced the number bars being used to generate strategies.

I was previously using 500,000 bars (M1), but I am now using 20,019 bars (H1) for 20 instances, and 7,653 bars (H1) for the other 20.

Interesting, if that is the cause of the difference.