Wednesday, October 05, 2022
Statcast Metric: Best Speed
One of the issues in using "average" exit velocity is that it treats a 60 mph and 100 mph batted ball the same as two 80 mph batted balls. The reality is that we learn nothing about a batter on their slow hit batted balls.
Math break: In order for an "average" to work, the impact of the gap between each number has to be roughly the same. If the gap in performance between 95 and 96 mph is the same as the gap between 96 and 97, then we can see that each unit of mph is worth the same. And therefore, it satisfied the conditions to average. When you look at the gap in performance between 70 and 85, you won't find that. This is why we really don't want to focus on poorly hit batted balls.
So, which batted balls are we interested in? We can approach this in a very specific way. We can focus on just the 5% hardest hit balls in one season and see how well we correlate to next season's performance. We can keep increasing our sample (10% hardest hit, then 15% hardest hit) and so on. When we do that for each step of 5% up until 100%, this is what we get (click to embiggen):
When we do that, we get a clear signal right close to the 50% mark. Repeating this for pitchers, and we get a similar result (though for pitchers, it's the 50% weakest batted balls).
And so, some time this week we will show on Savant, each batter's and each pitcher's average "Best Speed" batted balls. And focusing on the average of the 50% hardest hit for batters (and weakest hit for pitchers) is the better way to get at the talent level of the players, than looking at all the speed of all their batted balls.