Exploring Pitcher Ball In Play Allowed Data
I spent most of my free time this morning designing a new dashboard tab, so I just want to talk about that here today. The purpose of the new dashboard tab is to see data on pitcher ball-in-play performance.
Only paid subscribers have access to the dashboard, so become one today if you’re not already and want the full dashboard!
You know that the primary stat I look at with pitchers is K-BB%, because those are the things most in the pitcher’s control and they tell you a good amount about who the pitcher really is.
But what happens after contact is hugely important as well. The problem with this is that this stuff isn’t very sticky. If you perform very well in contact allowed for a season, that doesn’t tell you very much about what you’ll do next season - there’s a ton of randomness in it.
But since we know that it’s largely random, we can actually look at pitchers that are performing very poorly or very well in these categories and then expect regression toward the mean. That’s what this tab will do, and it’s very unique - I don’t think anybody else has even tried this to this point.
One thing we need to understand first is the launch_speed_angle categorization that Statcast uses. What that is is a classification algorithm that takes every batted ball and puts it into one of six categories based on the launch angle and launch speed. This is where barrels come from. Here is a quick plot of how they all line up:
So just to take an example, the barrels are orange. You can see they are all hit above 97 miles per hour, and they are at an angle range mostly between 15 and 40 degrees. As the velocity goes up, the angle range widens - so it’s a dynamic classification. I hope that makes it clear.
What I wanted to do with this dashboard is see how every pitcher is performing in terms of two things
The breakdown of the batted ball types they’ve allowed (what percent have been barrels… what percent have been solids… etc.)
The performance of each of these in AVG, wOBA, and HR%
To reiterate the key point, there is a lot of randomness going into this. If a pitcher is giving up a 14% Brl%, and the league average is 8%, it’s unlikely that he continues to give up that 14% Brl% (not impossible, but unlikely). And then it’s also true that if a pitcher is giving up a homer on 75% of his barrels, and the league average is 50%, that 75% is likely to come down as we move forward.
The home park plays a big part here. Barrels will go for homers at a much higher rate in Cincinnati than they do in Kansas City, so we have to keep that in mind. There’s also some difference between lefties and righties and pitchers with really good stuff as compared to really bad stuff, and stuff like that - but in general terms things should regress toward the mean.
Okay, now we can look at an example. Here are the benchmarks (league averages), firstly:
» Brl%: 8%
» Solid%: 6%
» Flare%: 24%
» Under%: 25%
» Topped%: 35%
» Weak%: 4%
So that’s the percent of the time that a ball put into play falls in each category.
Note that Under% and Topped% are largely in a pitcher’s control since those are essentially fly balls vs. ground balls. Alex Cobb will always show up with a high Topped%, for an example. Those two are largely controlled by the pitcher, but the other four are much less so controlled by the pitcher (but a high ground ball pitcher will probably end up with a lower Brl% and Solid% since those are fly balls, stuff like that - just keep all that in mind)
Now we can look at the performance of each batted ball type:
That’s data just from 2023. Probably HR/BIP on Barrels will come up a couple of points as they play more games in hot weather, but more or less those are numbers you can expect to stay pretty steady.
So now we can look at a pitcher example to really bring this all home. Let’s take Lance Lynn, who is having a disastrous season so far despite having a pretty strong K-BB%.
Here’s what you’ll see when you select his name from the pitcher dropdown:
Zooming in:
The barrel rate is higher than league average for Lynn, but it’s not crazy high and Lynn hasn’t been a ground-ball pitcher in his recent career so it’s not crazy to think that might stay around 10%.
The topped% is down, which again is expected since his arsenal is prone to giving up more fly balls than ground balls.
The flare rate is higher than league average but again, not much different. This is a pretty standard profile; nothing super weird is going on.
Now let’s look at the performance of each compared with the benchmarks:
What sticks out here is the 77% HR/BIP on barrels. Lynn plays in a pretty neutral park, so we shouldn’t expect a high or low value there, it just seems like he’s been unlucky there.
He’s also given up an .803 wOBA on the flares, much higher than the average .640 wOBA the league has generated. And those flares are his most commonly allowed batted ball type, so that’s been hurting him quite a bit.
You could say with relative confidence that he’s due for some positive regression in what is happening after he’s allowing balls in play. You can check FanGraphs and sure enough, he’s given up a .364 BABIP, well over his career average of .299.
The point of this post wasn’t to talk about Lance Lynn, it’s just to show the new dashboard and give an example on how to use it. You will find this all on the “Pitcher BIP Analysis” tab of the JonPGH MLB 2023 Dashboard, which paid subscribers of this blog have access to.
But I’ll give everybody a free sample, here’s a sample view you can use to play around with it.