I'm fairly new to sports analytics, but this year I'm working on a predictive model for the NCAA tournament. I was wondering if, in any of your experiences, it's a fair assumption to treat regular season games and playoff games the same when making predictions. (i.e. is it unwise to train a playoff prediction model using regular season games). On the one hand, it's still just a basketball game. On the other hand, there may be certain predictors that show up more in a winner take all tournament than in a less meaningful regular season game. Any input would be awesome!
In case anyone's wondering, yes this is the kaggle competition (but I'm just doing it for fun, so don't steer me in the wrong direction!
).