The Most Improved Changeup of the Second Half

I’ve been attempting recently to better understand changeups through the prism of spin. A relationship might exist, but it’s difficult to identify. I’m working on it — and, if I find anything, I’ll share it here. In the meantime, though, there still remain some dependable pitch-level metrics which can reveal the quality of a changeup — namely, drop, fade, and velocity difference. While it’s true that there are a multiple pathways to success for the changeup, those which are notable by these criteria also tend to be notable for their effectiveness. Movement, for example, is what allows Zack Greinke’s hard change to work. The velocity gap between the change and the fastball, meanwhile, becomes more important for those changeups which feature less of Greinke’s signature movement.

Over the course of this past season, a few changeups improved in these regards. We should take notice because, even in today’s era of spin, a nasty changepiece can really pull an arsenal together.

What I’ve done for this post is to compare all the league’s various changeups by assigning a z-score for improvement in the three aforementioned categories, so that I can sum up “improvement” into one number. That number just allows us to sort the group; more important are the actual improvements in shape and velocity.

Most Improved Changeups of the Second Half
Name Diff Vert Diff Horiz Diff Velo Diff SumZ
Tyler Chatwood 1.3 1.8 2.5 6.1
Juan Nicasio 4.3 0.4 1.8 5.6
Patrick Corbin 2.7 0.4 1.4 4.0
Jhoulys Chacin 1.8 0.5 1.5 3.7
Justin Verlander 1.8 -0.1 1.3 2.8
Chase Anderson 4.6 -0.3 -0.2 2.5
Drew Smyly 2.7 0.1 0.4 2.5
Jon Lester 1.2 0.6 0.6 2.3
Ivan Nova 0.5 1.7 -0.1 2.3
Mike Fiers 0.1 1.5 0.3 2.3
Diff Vert, Diff Horizontal = difference in changeup movement vertically and horizontally between the first half. Velo Diff is the difference in the gap in velocity between the fastball and the changeup.

Tyler Chatwood’s changeup has never been good, but it really looks like he’s tinkering with it, no matter which classification system you’re using. By PITCHf/x, he added nearly two inches of fade and over an inch of drop. Meanwhile, he began throwing the fastball harder and change slower, improving the gap between the two pitches by 2.5 mph, the biggest difference in the sample of 92.

That’s according to the generic PITCHf/x classifications from MLB, at least. According to Brooks Baseball’s data, the improvement was a little more muted, suggesting that some of the difference comes not from any improvement on Chatwood’s part, but simply from mis-diagnosing changeups. He still added an inch of fade there, a half inch of drop, and a half mile per hour of velocity gap. The problem is that he pushed that changeup all the way to… below average with respect to movement. The velocity differential was merely average. Still not a good pitch, even if it was better.

Juan Nicasio is another pitcher who’s been around a while and convinced us that his changeup was no good in the process. There’s probably some classification stuff going on here, too. Brooks has him adding a half inch of movement in the good directions, but coming up well short of the benchmarks you want to see from a good changeup.

Patrick Corbin! Give us hope that you’ve actually discovered something! According to PITCHf/x, the left-hander’s changeup exhibited signs of improvement. Once again, though, the findings here aren’t corroborated elsewhere. Over at Brooks, that difference in vertical movement is non-existent. He did add some velocity differential, almost a full mph, but the pitch still comes up well short of the various benchmarks for the average change. He trusted it less, and it got mediocre results.

You’re probably starting to notice a trend: a challenge in this sort of analysis is ensuring that the observed differences of a pitch aren’t merely the product of changing pitch-classification criteria. Justin Verlander represents a case study for this sort confusion. His changeup averages about 86 mph, his fastball about 94 mph. The result: his fastest changeups and his slower fastballs tend both to travel at about 90 mph. The Brooks data contradicts the PITCHf/x data elsewhere, too, suggesting that literally nothing changed about Chase Anderson’s changeup in 2016 and that Drew Smyly threw one changeup after July ended.

Jon Lester. Here’s real improvement, on a guy that had been going away from his changeup for a while. In the first year during which he added changeups year over year since 2012, the lefty also improved the movement and velocity on the pitch. Even Brooks has him adding a half inch of drop and almost more than a half tick on the gun. Seems pretty muted, but we can believe it more than these poorly classified changeups we’ve found so far. He also had, in August and September, the highest two-month changeup usage he’s had since 2013. Seven percent!

Ivan Nova may have improved his changeup, and quietly added more of them in Pittsburgh, but the change was only to get back to his career levels. And those levels weren’t impressive.

The last guy on the list may actually have the best argument for legitimate improvement on a pitch. Mike Fiers threw more changeups this year than he’s ever thrown, and it added fade and some gap as he marched into September. Let’s look at an average early-season change against an average late-season change.

In March, the more vertical change:

And then late in the season, after he dropped his release point a few inches and got more fade on the pitch:

Legitimate change, but minor, and not super visually impressive.

If we focus on the velocity differential, the 15th-ranked improver according to PITCHf/x — Brandon Finnegan — jumps into the top seven. And that looks legitimate, as all systems say he added almost a mile and a half of velocity gap between the first and second halves. Teammate Dan Straily, who himself had to try 17 different changeup grips before he found the right one, showed Finnegan a grip that worked for him, and the lefty’s confidence in the pitch skyrocketed. He went from throwing the pitch around 10% of the time to 15%, and finished the season with a string of six starts with 20% changeup usage.

In that six-game stretch, he had a 2.37 ERA backed by a 30% strikeout rate. Sure, his 10% walk rate diminishes his upside, but a 10-mph gap on the change means that the lackluster movement is not as important now. Ten miles an hour is a benchmark, because it’s around that sort of gap that you get swings like Finnegan got on this changeup in September.

So, sure, Mike Fiers improved his changeup the most by incrementally improving in all three aspects as he threw the changeup more and more often over the course of the season. That’s not sexy, but it’s a finding. It’s also interesting that a young lefty in Cincinnati made huge strides with his changeup, as well, and it might be a bigger deal going forward. Brandon Finnegan always had that plus slider, after all, and above-average velocity for a lefty. That change could mean everything.

With a phone full of pictures of pitchers' fingers, strange beers, and his two toddler sons, Eno Sarris can be found at the ballpark or a brewery most days. Read him here, writing about the A's or Giants at The Athletic, or about beer at October. Follow him on Twitter @enosarris if you can handle the sandwiches and inanity.

newest oldest most voted

Any way to get home/road splits on Chatwood’s change?

I mean, we need a way to explain THIS 2016 leaderboard: