Probably, better safe than sorry.

Run 1000 times against the artifical price curve of the previous unseen data, or against the artificial curve of the previously used train data.

Or again both?

Last edited by danatrader; 05/31/20 09:03.