Many thanks to you both for your input – I find the statistical theory behind all this as interesting as the code itself.
We would need to ensure we re-optimise at the 10K bars point until 100K bars have passed and then compare results of the 2 Versions (10K and 100K opti).
I was thinking something the same, Forward Testing but starting in the past: eg optimize Q1 of 2018 then run it over Q2; optimize Q2 and run it over Q3 etc.
Or is that more or less what you would get with your System 1: Walk Forward over 10K bars Not-Anchored ?
