Ryan: “[H]is split-change is pretty nasty, and he really improved his fastball command last year”
Me: “That split is soo sexy. His fastball is plus by every measure, too: spin, velo, pop-ups...I’m wondering if he might benefit from throwing more splits. Not as many as Shoemaker, but up into the 25% range.”
Ryan: “Now there’s a good article!”
I’m not going to turn down a pre-approved article idea, so without further adieu, let’s go!
Heavier reliance on “secondary” pitches is becoming much more commonplace. Starting pitchers like Rich Hill (49.7 percent curveball), Chris Archer (40.3 percent slider), and Matt Shoemaker (40.0 percent split-change) are testing the boundaries of how often non-fastballs can be thrown, and as far as I know, the point at which the usage of these pitches becomes detrimental to their effectiveness has not been breached.
Matt Shoemaker provides an interesting and relevant case. For his career, prior to May 16, 2016, he’d used his split-change 21.6 percent of the time. From that day forward he went back to that well constantly, using it an astonishing 40.4 percent of the time for the rest of the season.
For the record, no starter had ever eclipsed 40 percent usage with their split-change before, so Shoemaker was really taking a risk – but it paid off. While the sample size is still small, Shoemaker’s career numbers through May 11 pale in comparison to the 21 starts after.
|Career up to 5.11.16||301||4.13||4.09||15.2||39.3||2.01|
|5.16.16 - 9.4.16||135 1/3||2.93||3.01||19.6||41.8||1.55|
Gausman could learn something from Shoemaker’s audacity. Gausman’s own split-change compiles results that compare favorably not only to the league averages for splitters and change-ups (minimum 200 pitches), but even looks at home among the most used split-changes in the game - superior, even, to Shoemaker’s.
|Alex Cobb (2014)||22.9||62.4||N/A||38.1|
Among all splitters and change-ups thrown at least 200 times, starting pitchers only, Gausman’s split-change induced the ninth highest rate of whiffs and the third highest rate of whiffs/swing (out of 110 pitchers). Compared only to splitters, no one was above him in either category. Those numbers are even better than what Koji Uehara’s split earns, and he throws his 45.2 percent of the time.
It doesn’t stop there. Gausman’s average exit velocity ranks twenty-second among all splitters and changeups. It’s a nasty, sexy pitch.
If it were really so simple, Gausman would just throw his split-change more. The splitter is a feel pitch, however, and the ability to throw it well waxes and wanes. There’s also the issue of his repertoire. Gausman is equipped with a tremendous fastball - it featured the seventh highest average velocity of any qualified starter in 2016! It’s unlike any of the pitchers’ fastballs in the above group, and with its plus spin rate it generates above-average whiffs, pop-ups, and ground balls, somehow. You don’t just shove that pitch to the back seat.
|Velocity||Whiff||Spin||Exit Velocity||Pop Up%||GB%|
|Alex Cobb (2014)||91.5||5.0||N/A||N/A||7.0||47.6|
On the opposite end of the spectrum, the main knock on Gausman is that he has yet to harness a breaking ball, which is generally thought of as the pitch a pitcher needs in order to get same-handed batters out. This doesn’t imply a lack of effort, however, as the pitch has clearly undergone a remodel.
|2016||Horizontal Movement||Vertical Movement||Velocity|
|Rt. Handed Sliders||2.5||2.1||84.3|
|Rt. Handed Curveballs||6.2||-5.7||76.3|
That remodel led to a classification divergence among baseball websites. Brooks Baseball classifies his 2016 breaking ball as a curveball while Baseball Savant deems it a slider. Looking at the characteristics of Gausman’s breaking ball, we see a different shape in 2015. The truth is that it moves like a slurve, but produces results akin to a curveball.
On all breaking ball types, the league whiffs 15.5 percent of the time, so by that measurement, Gausman’s breaking ball lags. Of the pitchers who threw their split-changes the most, however, all get average whiffs on their breaking ball, though it should be noted that Shoemaker and Masahiro Tanaka both eclipse the 15.5-percent whiff rate for breaking balls with their sliders. In other words, Gausman fits a type.
|Pitch Type||Usage||Whiff%||GB%||Exit Velocity|
|Alex Cobb (2014)||Curveball||20.0||10.4||63.4||N/A|
The point of talking about his breaking ball is to show that he has a third, average pitch—something to get right-handed hitters out—which he supposedly did not have until last season.
I say “supposedly” because of another encouraging sign. Not only has Gausman increased the utilization of his split-change against same-handed hitters, but the results have been extraordinary—29.7-percent whiff rate, and a 55-percent ground ball rate. Those figures are great on their own, but the fact that it could take some pressure off of his slurve is an added bonus. It’s an elite pitch, and holding it back against same-handed hitters, based on its results, would be foolish.
The bottom line is that Kevin Gausman is a minor tweak away from taking that next step, something that could be said of a lot of major league players. Not all cases raise the prescription that I propose for Gausman: Throwing more split-changes might be a key to his success.
I’m not sure if this what Ryan had in mind when he said this was a good article, but here’s what I want to do: I want to go into Gausman’s numbers from 2016 and see how much throwing more split-changes would have benefited his overall line. Herewith:
Disclaimer: I know this isn’t going to be perfect, and there’s no way it could be without a time machine, but it is fun. I also want to keep it as realistic as possible, so every factor that we can keep as a constant, we will. I’ll also pepper in gifs of Gausman’s split-change because...
This is the foundation for the entire exercise, as we’re testing what would have happened if he threw more split-changes. Gausman’s identity is “power pitcher”, and in 2016 he relied on his fastball 65.4 percent of the time. Because of this, we’re not going to send his split-change usage into the Shoemaker stratosphere—let’s keep this somewhat realistic—so instead, we’ll work in Tanaka’s range. Here’s the mix I’m proposing.
|Fastball%||Raw count||Curveball%||Raw count||Split-change||Raw count|
Among its many traits, Gausman’s fastball also serves as his most reliable pitch when he needs a strike.
His walk rate may suffer, since his raw fastball usage drops by 231 and we’re replacing those fastballs with pitches he throws for balls more often. Ultimately, we find that this new pitch mix nudges his strike-to-ball ratio down from 1.80 to 1.74.
Once we understand how well ball-throwing frequency correlates with walk rate, and how Gausman performed relative to his expected walk rate, we can determine his new BB%.
That’s a pretty decent correlation. Now based on the equation for the trendline: BB rate = (.728*percent of balls thrown)-.188; we find that, in 2016, Gausman only walked 86.4 percent of his expected walk rate of 7.2. His new ball-to-strike ratio would’ve produced a 7.7 walk rate, but allowing for the same percent of “overperformance”, we can adjust that mark to 6.6.
The goal here is to generate enough whiffs to counteract the uptick in walk rate, and ideally increase his overall K-BB performance. The new pitch mix we’ve created logs a 13.3-percent swinging strike rate (foul tips included on Baseball Savant) which compares favorably to his actual mark of 12.3. We’ll run the same analysis we used for walk rate by incorporating the correlation between swinging strike rate and K rate.
Gausman’s actual 23-percent K rate falls short of the 25.7 mark we could’ve expected from him based on the trend line equation of: K percent = (1.925*swinging strike rate) + 0.02. So we’ll have to adjust the expected 27.5 rate down to 24.5.
Oh yeah, and...
Batted ball distribution
To determine batted ball opportunities, we subtract walks, strikeouts, and hit batsmen from his 757 total batters faced. With his new pitch mix, he does hit one more batter, so the difference is 15 fewer chances for a batted ball. In this scenario, overall batted ball events drop from 531 to 516.
Things get a little complicated because if we want to keep the ratio of batted balls per pitch type the same as the actual outcomes Gausman experienced in 2016, we end up with too many batted balls—seven, to be exact.
|Pitch count||Batted Balls||Batted Ball%||Revised pitch count||Batted Balls||Batted Ball%|
This was never going to add up perfectly, since I can’t adjust pitch total or batters faced—numbers that probably would have changed given his new K-BB profile. There’s no precise way to account for what pitches wouldn’t have been batted. So for the purpose of this exercise, I’m going to subtract two fastballs, two curveballs, and three split-changes.
To determine Gausman’s revised batted ball profile, now all we have to do is apply the actual batted ball type rates to the quantity of times each pitch was put into play.
His revised batted ball profile:
Looking at the new batted ball profile, we can already infer that Gausman would have suffered fewer home runs simply because the decrease in overall fly ball and line drive rates would have taken away opportunities for long balls. However, we’re keeping the HR/FB and HR/LD rates constant based on pitch type, and his split-change, because of the sample size, allowed the highest home run rate. As it shakes out, Gausman allows two fewer home runs overall.
|HR/FB||Fly ball count||Home Runs||HR/FB%||Fly ball count revised||Home Runs revised|
|HR/LD||Line drive count||Home runs||HR/LD%||Line drive count revised||Home Runs revised|
I feel like it’s time for one of these...
To determine how many batted balls can actually be classified as “balls in play”, we subtract the new home run total, 26, and errors in the field while Gausman was on the mound, seven (was not affected by the drop in opportunities), from the number of batted ball opportunities we came up with earlier, 516. If you’re a human calculator, you’ve already figured out that there would have been 483 balls in play against our Frankensteined Gausman.
Using fly balls hit off his fastball as an example again, and after accounting for home runs and errors, this specific BIP type yielded a .068 BABIP, a .236 AVG, and an .819 SLG.
Here’s how all the BIP types shake out:
|FB||FB BIP||BABIP on FB||H||2B/H||2B||3B/H||3B|
|GB||GB BIP||BABIP on GB||H||2B/H||2B||3B/H||3B|
|PU||PU BIP||BABIP ON PU||H||2B/H||2B||3B/H||3B|
Now we can start turning this into runs! From FanGraphs’ glossary:
Weighted On-Base Average combines all the different aspects of hitting into one metric, weighting each of them in proportion to their actual run value. While batting average, on-base percentage, and slugging percentage fall short in accuracy and scope, wOBA measures and captures offensive value more accurately and comprehensively.
Last year, the formula for wOBA was:
We have all the information we need to come up with Gausman’s new wOBA allowed.
His wOBA against would’ve been .309 compared to .319 it was last year. So what does that mean for his ERA?
wOBA correlates very well with ERA, although Gausman was a pretty big outlier last year, as his 3.61 ERA came in well below the expected 4.20. A .309 wOBA correlates to an expected 3.87 ERA, but adjusted for what happened in 2016, we can knock his ERA down to 3.31.
Who doesn't want to see more of this?
It’s possible that the former consensus top 35 prospect, and number 10 prospect via Baseball Prospectus, continues to improve via the more ordinary road of refining his third pitch, his breaker, and I think that’s a fine idea, too. But as we see more pitchers conclude that their best weapon doesn’t necessarily have to be a secret, opposite-handed hitter, two-strike weapon, it’d be an oversight to cap the use of his split-change to 21.3 percent. Gausman has improved every year since coming into the league and I believe he’s capable of making another leap this year. I can feel the wind of split-change.