How Did Madison Bumgarner Fix His Curve?

The thing about the curveball is getting batters to swing. Once you get the batter to swing at your curveball, it has the same whiff rates, basically, as a changeup or a slider, especially once you correct for the fact that the curve is the slowest pitch type, meaning batters have an easier time making contact with it. But the swing rate against the curve? Easily the lowest in the game — below 40% when most other pitch types are near 50%.

If the swing is the thing generally, then it’s no surprise that getting batters to swing at his curveball has been a major part of Madison Bumgarner’s excellent season after a less-than-excellent first month. He admitted as much when I talked to him in May: “I just don’t feel quite right yet,” he said then. “They haven’t been swinging as much at my curve.”

In 2015, Bumgarner recorded a 46% swing rate on his curve, good enough for fifth-best among starters who threw their curveballs at least 500 times (21 total). Despite seeing a lot of curves, and despite the fact that the big lefty is not a leader in curveball zone rate, batters were swinging at the pitch. Then? April 2016 happened.

Since Bumgarner pushed his strikeout percentage north of 25% for the first time in 2014, he’d never seen a swing rate so low on his curve as he saw in the first month of 2016. When I asked him why, Bumgarner said “it isn’t as hard” and that he was looking at mechanics, and “something earlier than” what happened at his hand break.

The first part is easily testable, and seems like an obvious answer to the titular question. Why weren’t they swinging at his curve? Because it was slower than it had ever been.

But we haven’t solved this situation completely. The irascible lefty has pushed his curveball velocity back up this year, but not to where it was before. And yet he’s gotten his swing percentage on the pitch north of 50% in two of the last three months, which is vintage ‘Garner.

Unfortunately, when I approached him late in the season, he wasn’t about to reveal any specifics on the pitch. “I found something in my mechanics,” he said coyly before declining to elaborate.

Interestingly, it’s not about release point. His vertical release point on the curve is a bit different this year, but it hasn’t changed over the course of the season. His horizontal release point on the pitch is about the same as last year. The vertical movement on the pitch is the same as ever. His two-strike usage of the curve is the same last year as this year (26%), as is his two-ball usage (17%).

There is one thing that looked weird earlier this year and then rectified itself along with the velocity: the horizontal movement on his curve.

Look:

That graph mirrors the velocity and swing-rate graphs, and it suggests that perhaps he was getting around the ball too much early on. He was getting more horizontal movement but less velocity. Unfortunately, that sort of thing is be tough to see in a moving image form, even if we find a “typical” curveball from the early season to compare to the late season. Let’s try, anyway.

Early (73 mph, -7 inches horizontal movement):

Late (75 mph, -5.8 inches horizontal movement):

Hard to see much difference, and so we’re left guessing. My guess is a predictable one, given my interest in the subject: I think he fiddled with the grip. That could change how much horizontal movement he got on the pitch, and help him stay behind it to get a few more mph on the pitch. It would also be a mechanical change that came before the hand break.

We don’t know exactly how Bumgarner got his curve back, but we know it has something to do with velocity and movement. If he’s throwing 75-77 on the curve in tonight’s game, it’ll be a good sign for him and for his yakker.