Please beware that, the tool was just to see the consequences of any hypothetical portfolios. Finding optimized parameters is inadvisable - for they are unlikely to hold up in the future, unless they make fundamental sense.
Now I'm working on the more important component of this Darwinex project - Finding good darwins. With this, we could construct hypothesis as to what makes a good darwin portfolio, find fitting darwins IN THE PAST (and while hypothetically remaining blind to their future performance) and then looking at their future performance through portfolio simulator - to see if the hypothesis holds up.
This will hopefully if not remove, then reduce overfitting. Though I'm juggling a few too many projects at the same time, so it's slow going.