Quote:
Originally Posted by Carter08
Wins in baseball don’t matter. I’ve “learned” that.
|
Wins in baseball is everything. Wins as a statistic attributed to a pitcher don't matter. There's a difference.
When I'm building a statistical model, I always use wins as the target variable. I want to know which team is most likely to win the game and what their odds of winning are. But far, the most important factor in that model is who the starting pitchers are. I can throw 100+ variables at the model and the mathematics will determine which of them are important. But it turns out that if you want to predict wins, you should actually look at other statistics besides "wins" attributed to a pitcher, because they are simply irrelevant in the presence of other variables like SIERA, K/BB, xFIP, velocity, WHIP, or even ERA. Knowing a pitcher's win/loss record literally adds zero information mathematically speaking. It is a nothing burger that only serves as a surrogate for things outside of one's control. I can't imagine a noisier statistic to look at.