From a strickly "Statistician" point of view, I'd want the autumns of 2007 and 2015 explained first, and then I'd investigate why there are no material drawdowns. Although there are statements in the verbage regarding robust backtesting and estimation that's not dependent on immediate-past success, the *shape* of that profit curve begs, "Everyone's a genius in a bull market." (And those curves even predate March'09!)
Nice. I use a few AI systems, and they work if modeled correctly. Definitely not for the beginner trader.
I got this up and running on my machine, and so far, I got nothing, not a single trade. But its cool, teaching me a bit about Java, and I will keep playing with it.