The Second Half’s Most Improved Changeup
As much as certain aspects of pitching remain a mystery when it comes to analytics, we’ve figured out a few things about the changeup. Like, movement is good! And like, a bigger velocity gap is good if you want whiffs! Those sorts of relationships can be identified pretty easily. And since movement and velocity become stable really quickly, you can cut this sort of thing into smaller samples and still get meaningful results.
You can ask things like: whose changeup improved the most in the second half, when it comes to things like velocity and movement differential off the fastball? And then you can get answers.
The answer to that particular question appears below in the form of a table. A detailed description of the data appears below the table, so you can click to see it. In short, I isolated changeup movement and velocity off of the fastball in both halves of the season and then compared second-half scores to first-half ones.
Here are the top-10 changeups in baseball by overall second-half improvement:
| Player | Changeups | 2nd Half CH Whiff% | 2nd Half CH GB% | 1st Half Z | 2nd Half Z | Diff Z |
|---|---|---|---|---|---|---|
| Dallas Keuchel | 214 | 25% | 75% | 1.23 | 3.82 | 2.59 |
| Lance Lynn | 36 | 0% | 44% | -1.08 | 1.00 | 2.08 |
| Robert Gsellman | 114 | 13% | 72% | -1.30 | 0.64 | 1.95 |
| Fernando Abad | 76 | 9% | 62% | -1.83 | 0.04 | 1.87 |
| T. Hildenberger | 228 | 22% | 85% | 3.15 | 4.81 | 1.66 |
| Mike Leake | 178 | 6% | 83% | 0.29 | 1.92 | 1.62 |
| Marco Gonzales | 321 | 14% | 65% | -1.41 | 0.18 | 1.59 |
| Jake Arrieta | 100 | 17% | 67% | 0.65 | 2.20 | 1.55 |
| Blake Treinen | 53 | 25% | 71% | -0.24 | 1.29 | 1.53 |
| Enny Romero | 46 | 17% | 67% | -3.08 | -1.58 | 1.49 |
| Felix Pena | 48 | 19% | 100% | -0.50 | 0.97 | 1.46 |
Dallas Keuchel has always had a great change. The real development for him this past year was the four-seamer. It went from having four inches of fade to having none in the second half — almost like a cut fastball instead of a four-seamer — and that really fueled the difference in the changeup. The change itself didn’t see its movement altered much, though.
Trevor Hildenberger’s changeup got better in the second half, which probably helped him record one of the top-25 strikeout- and walk-rate differentials (K-BB%) over the latter half of the season. Mike Leake and Jake Arrieta saw their changeups improve from meh to better.
But maybe most impressive was the improvement in Lance Lynn’s changeup. It’s in a small sample, yes, but movement and velocity become meaningful very quickly. He’s been around for a while and has never had a good changeup. In the second half, though, he made incremental improvements on the pitch and it flashed potential.
Since the pitch itself changed — as opposed to the circumstances around it — and since Lynn is a free agent, this seems like a good time to take a deeper look. Let’s inspect how his movement and velo looked in both halves, set against the league average for right-handers in 2017.
| Player | Horizontal Move | Vertical Move | Velo Gap |
|---|---|---|---|
| Lynn First Half | 1.9 | 3.4 | 7.7 |
| MLB Average Righty | 2.6 | 4.7 | 9.0 |
| Lynn Second Half | 3.0 | 4.1 | 8.2 |
As you can see, the changeup improved in all aspects. It was a gradual development, but the evidence is there.
Let’s look at pictures! Here’s a totally normal first-half changeup.
And now a second-half changeup that represents the new movement and velo well.
Obviously, this isn’t an ideally located pitch, but there’s more movement than the previous version of the change and it’s slower. Here’s the caveat, though: as is his wont, Lynn didn’t throw the changeup much (3% for the year), and threw it even less in the second half (2%). He didn’t get a whiff on the changeup in the second half, either. He loves to throw his fastballs. In the second half, he threw his cut, four-seam, and two-seam fastballs a whopping 93% of the time.
Lynn manipulates his fastball really well — Jeff Sullivan did a great deep dive on that ability — and the skill is part of why he’s been good despite lower strikeout rates. He’ll probably continue to vary his fastballs a lot with his new team — as Max Weinstein found in a deep dive on fastball mixers — but this changeup, and his cutter, offer a bit of upside beyond.
What if he mixed his fastballs for weak contact and called strikes, and also threw this better changeup and cutter more often for swinging strikes? As he continues to distance himself from a Tommy John procedure, his command should improve, as well, which could filter down into his non-fastballs as well.
The projections don’t paint a rosy picture for Lance Lynn’s 2018 season, so he might not cost much to his acquiring team. But not only is there an existing fastball-mixing skill that’s not currently captured by projections in his current arsenal, there’s potential for an emerging new pitch in there as well. Lynn’s an intriguing watch, and signing, for a guy who mostly throws fastballs.
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.

I thought this was going to be about Robert Gsellman after first glancing at that table.
FWIW, unadulterated Statcast (which of course isn’t perfect) says otherwise re: velo:
First half: Average changeup velo: 85.6 MPH. 87 changeups (4.9% of 1758 pitches). 5 whiffs, 12 called strikes, 5 fouls, 53 balls and 12 balls in play. The balls in play were hit at an average of 89.5 MPH (not good) at an average angle of 24.2 degrees.
Slash line on Lynn’s 13 PAs ending with a changeup, first half: .417/.462/.917
Second half: Average changeup velo: 86.3 MPH. 79 changeups (or 5.6% of 1411 total pitches) for Lynn. 4 whiffs, 15 called strikes, 8 fouls, 40 balls and 12 balls in play. The balls in play were hit at a very soft average of 82.3 MPH at an average angle of 16.9 degrees.
Slash line on Lynn’s 13 PAs ending with a changeup, second half: .333/.385/.833
A question this raises is whether the chart above re: velo gap should show the league average gap in the first half and league average gap in the second half? Wouldn’t pitchers’ increasing fastball velo on average as the year goes on impact the average velo gap as well?
I think part of the problem with this overall analysis with Lynn is that he tends to manipulate his fastballs, as the post pointed out. I’ve seen him just drop a 2-seamer in the zone on the first pitch with a little taken off many times and it’s usually classified as a changeup, at least on the ESPN play-by-play.
That for sure makes sense.
Max velo changeup for Lynn in 2017, per Statcast: 89.8 mph
Min velo changeup for Lynn in 2017, per Statcast: 80.7 mph
Very much prefer Pitch Info’s hand-checked classifications.
Hey eno, where can I find average vertical/horizontal movement for pitches at?
I’m not sure this is what you’re looking for exactly, but here on Fangraphs. Go to a pitcher’s page, go to “Splits”, then “Pitch Type Splits”.