With the technological ability to track everything that happens in a ballpark during a game rapidly improving, and also the ability to collect, dissect, and present that information in useful ways similarly improving, it was always likely to be the case that we’d see another step forward from Statcast this season.
After all, baseball statistics need not include only the numbers recorded in the box score (and the derivatives thereof) – why can’t they also be derived from the ever-expanding pool of data being collected, literally, by high-definition cameras and Doppler radars in every ballpark?
[adinserter block=”1″]
Sure enough, in conjunction with this weekend’s Sloan Analytics Conference, the MLB.com and Statcast crew announced two new stats that will be presented this season: Catch Probability and Hit Probability. The short version on those is just about what you’d expect: what is the probability that a ball hit like that is caught and/or lands for a hit? In so analyzing, we get greater context for what our eyes sometimes fail to accurate tell us.
For example, this is how Albert Almora’s brilliant diving catch in the NLDS against the Giants could be presented with the new analysis:
Your browser does not support iframes.
So, then, based on the distanced needed to be covered and the time an outfielder had to get to that spot, there was a 21% chance that a ball like that is caught. Sounds about right for Almora. (Also, the catch was all the more impressive given that it prevented a Giants walk-off win right there in that moment. That’s the kind of thing for which we’ll have to continue to provide narrative context.)
[adinserter block=”2″]
By contrast, check out this Matt Kemp diving catch:
Your browser does not support iframes.
Crazy awesome catch, right? Well, I don’t want to take away too much from it, because he did make a nice play on the ball. But Statcast can determine that the ball actually had a whopping 75% Catch Probability – it’s usually caught. Knowing that, and then watching the play again, you can more easily see how that’s the case: Kemp’s break was not swift, he’s no longer especially quick, and he took a poor route to the ball. When you observe the flight path of the ball, itself, it just looks like a ball that should be caught.
[adinserter block=”3″]
For batted balls – Hit Probability – Statcast uses our familiar friends exit velocity and launch angle. Not only is that a stat that could be fun in the moment, it can allow you to do more involved analysis. For example, Mike Petriello dug into which 2016 starting pitchers would have the lowest expected opponent OPS, based on non-batted ball elements (walks, strikeouts, etc.) as well as Hit Probability – and sure enough, Kyle Hendricks and Jon Lester both show up on the list. That’s especially fun to know, because it is completely and totally exclusive of team defense.
Read up on the new stats here and here, as well as an excellent write-up on the coming changes for baseball and statistical technology from Jeff Passan.
I look forward to seeing how these (and increasing permutations) are used this season by MLB-proper, and by those of us on the outside looking to dig in more deeply on discrete topics throughout a given season.