I am in the middle of my investigation so I have no proper conclusions as of yet (and probably never will).
But it is appearing to me that when evaluating 1000's of strategies and making selections among them if based on draw reward only it will pick the ones that work least out of sample. It is like a local optimum get reached more swiftly. This would make sense as there is an inefficiency developing and is closed quickly by other market participants. This would also explain why most strategies cease to last for ever.
When looking at...Ignored