I think your point is correct but you are not understanding that correlation does not imply causalitt
But how large of a historical data set do I need to show before it can be considered causality? At some point one has to wonder maybe there is something there. Not at the individual player level of course. Too much noise in that. However, I think it could help buck the notion that you have regress all players BABIP to 300.