Using TrackMan and Pitch f/x Data to Grade Curveballs

Who has the best curveball? Does Seth Lugo’s league-leading 3300 spin rate make his the best? Or does Mike Fiers’ 11 inches of vertical break make his curveball superior? Or how about Aaron Nola’s 10 inches of horizontal break?

While it’s fun to compare and rank pitchers based on spin rate and movement, it’s hard to know which pitch-tracking measurements most affect the quality of a pitch. Sure, it’s nice to say your curveball spins the fastest– but what does that really mean?

Baseball Prospectus has a statistic called True Average (TAv) which incorporates an offensive players entire performance and scales it to batting average. TAv is similar to FanGraphs’ wOBA in its use of linear weights, but it also goes a step further and accounts for ballpark and league effects. I suspect that TAv can be used for pitchers too. TAv against a particular pitch (e.g. curveballs) is a good indicator of how batters perform against it.

But what are the factors that most influence a batter’s TAv against a curveball? Using data from 2015 and 2016, I regressed pitcher’s curveball velocity, spin rate, horizontal movement, vertical movement, and a dummy variable to test for a difference between pitcher handedness, on their TAv. I would have liked to include more seasons but spin rate data was only available since 2015. Anyways, here are the regression coefficients I found:

Variable OLS
Spin Rate -3.29E-06
Velocity -0.005***
Horz. Movement -0.003
Vert. Movement 0.004**
RHP 0.031***
Constant 0.612***
Observations 270
R-squared 0.09
***p<0.01, **p<0.05, *p<0.10

Essentially, what this means is that a higher velocity and a greater vertical movement make a curveball more effective, albeit only slightly. For every one additional mph in velocity, a batter’s TAv against a curveball decreases by 5 points. Additionally, an increase in one inch of vertical break decreases TAv by 3 points. The dummy variable RHP shows that, on average, left-handed pitchers have more effective curveballs than right-handers.

The coefficients for spin rate and horizontal movement were insignificant. At least in this model, those two measurements do not impact the effectiveness of a curveball.

Using my model, here are the pitchers who have had the best curveballs since 2015 (with at least 100 curveballs thrown in each of 2015 and 2016):

LHP RHP
Player xTAv Player  xTAv
Drew Pomeranz 0.166 Lance McCullers 0.177
Matt Moore 0.168 Sonny Gray 0.184
Gio Gonzalez 0.171 Jake Arrieta 0.185
Rich Hill 0.178 Stephen Strasburg 0.187
James Paxton 0.179 Jimmy Nelson 0.187

 

At first glance, this list seems to pass the eye-test. Pomeranz, Hill, and McCullers are more or less known for their curve– all throw the pitch more than 35% of the time. McCullers was acquired off waivers in 2014 because his curve showed an atypically high spin rate (there goes the validity of my research). Nelson (14% curve usage, 4.37 ERA) and Moore (21% curve usage,  4.41 ERA) are two who have struggled a bit over the past two seasons. Maybe it’s time for them to start throwing some more curveballs.

Posted by Jeb

From the great "vacationland" state of Maine. Former D3 baseball player on an underachieving team. Prior: TrackMan with A's. Current: Check Down Sports. Soon: Video with Reds. All-time facial hair lover.