When to Use Current BvP
So with the lack of predictive of ability of Batter vs Pitcher data, you might think I never even look at it. That’s not actually true.
Thus far, I’ve analyzed how useful BvP is on the individual level. And the answer is…not all that useful. But, in aggregate, I think the numbers can have some value. Remember, our problem with BvP shouldn’t be that it doesn’t “exist” or that it’s a total illusion or that there isn’t any predictive value, but rather with the sample sizes involved with analyzing one batter against one pitcher. Lots of times we might be getting a signal, but the problem is there’s so much noise that we can’t differentiate the two; the chances of the 5-for-10 hitters really “owning” the opposing pitchers are just barely above what we’d expect to see from randomness—and BvP has historically led to less value than simple platoon splits—that we just can’t trust that what we’re seeing is real.
I think if we’re going to make BvP useful, we obviously need to find ways to overcome the small samples. We’re basically never going to see a batter who has enough at-bats against a pitcher to really put a ton of confidence in whatever numbers we see. So before I propose a revised form of individual BvP in the following lesson, I want to talk about ways I think current BvP can be useful—if we employ it on a less granular level.
To read the rest of this lesson, you must purchase the course!
