How Predictive Is Expected Home Run Rate?
Last week, I dug up an old concept: expected home run rate. The idea is deceptively simple: assign some probability of a home run to each ball a batter hits in the air, then add them up. It tells you some obvious things — Fernando Tatis Jr. hits a lot of baseballs very hard — and some less obvious things — before getting injured, Aaron Judge had lost some pop.
One question that many readers raised — reasonably so! — is whether this expected home run rate actually means anything. The list of over-performing hitters was full of sluggers. How good is this statistic if it tells you that good home run hitters are, in fact, not as good as their home runs? Sounds like a bunch of nonsense to me.
In search of truth — and, let’s be honest, article topics — I decided to do a little digging. Specifically, I wanted to test three things. First, how stable is expected home run rate? In other words, if a player has a high expected home run rate in a given sample, should we expect them to keep doing it? If the statistic isn’t stable, what’s the point?
Second, how does it do at predicting future home runs? In other words, does an expected home run rate in, say, July predict what will happen the rest of the year? It’s also useful here to see if expected home run rate (from here on in, I’ll be calling this xHR% for brevity) outperforms actual home run rate as a predictor. If xHR% doesn’t do a better job of explaining future home runs than actual home runs, what use is it? Read the rest of this entry »