Checking in on Pythagoras

I tried to come up with a few ways to predict future outcomes better than the tried-and-true run differential we love so much.