So I trained until 1/1/2019 and put that into a ranking system called AI. The sim is after that: starting 1/1/2019 with that ranking system.
We can do that now or did I make a mistake?
BTW, out-of-sample 31 CAGR is looking good for some excellent Designer Models. Did @Marco succeeded in reducing overfitting to some degree or at least give us a method that gives better expectations for the out-of-sample results than a simple backtest would?
Jim