A Different Way to Look at Sliders

Talking to Mets pitching coach Dan Warthen last year, he mentioned something about the strategy of his brand of slider that has stuck with me. We normally think of sliders as hard curveballs, maybe. Loopy but hard pitches. Try figuring out if Craig Kimbrel throws a hard curve or a slider, and you’re down that normal path.

But Warthen said something a little different about the slider: “We don’t want to make it break, we want to think about getting our fingers to the front of the ball and spinning the baseball. Then you take another breaking ball and you separate the speeds, and it doesn’t have to be a great breaking ball, it just has to be a different speed.”

So, in effect, Warthen was talking about changing speeds with the slider. Normally that’s something you talk about with the changeup, which is obvious because of the way the pitch is named. But now we can talk about it with respect to the slider.

Let’s look at this year’s sliders (minimum 100 thrown) through this lens. In the graph below, you’ll find each breaker represented in four ways. The horizontal and vertical movement are shown on the x- and y-axis. The size of the mark is the difference between the slider and the fastball in mph, where the bigger bubbles have a bigger differential. And the color of the mark is the speed of the breaker by itself.

In general, you’ll find that the faster sliders feature a smaller velocity differential from the same pitcher’s fastball. They also move less, so you’ll find them closer to the y-axis and generally above the x-axis. And yeah, Jacob deGrom is the patron saint of this phenomenon, with a 90.5 mph slider with a (1,4) x-y value and a 4.4 mph velocity differential. He’s got a bright red little bubble.

Once you start talking about big green bubbles, though, things get dicey. They’re all over the place. But you’ll see lefties on the left and righties on the right, and generally the idea is that you get more movement in all directions once you slow it down. In a way, Kendall Graveman (74 mph slider, 16.4 mph differential, 8x, 1y) and Oliver Perez (77, 18, -6x, 0y) are the patron saints of the slow, big slider.

And then look at the outliers. There’s a delicious one. Down in the land of green, go find Felix Hernandez. He’s got movement like a Cam Bedrosian slider, or a Tim Lincecum slider. But his slider goes 85 and sits about 7.5 mph slower than his fastball. So it’s probably hard to pick up by velocity, and then it drops like a hard curve. Sonny Gray is a little like this, as his new slider is hard (87 mph), close to his fastball (6 mph), but has six inches of glove side movement and sits just below the slider thrown by Tony Zych, who has very different velocity stats but similar movement.

So it looks like some of the best sliders out there are just… different. There is an element of interaction with the fastball, at least.

Take the 10% with the smallest fastball differential, and their sliders have an average 13.9% whiff rate and a 46% ground-ball rate. The 10% with the biggest fastball differential have a 16.6% whiff rate and a 43% ground-ball rate. That’s just by differential, and it’s just bucketing, but it suggests that the closer sliders are maybe better for ground balls.

We can — or should I say, Matt Dennewitz can — run weighted regressions against whiffs and grounders where we weight the movement, velocity, and velocity differential by the number of sliders thrown. That’s the best way to isolate the relative importance of each of these variables.

Aspects of Sliders Related to Outcomes
 Target Stat R-squared X-Mov Y-Mov Velocity Velocity Differential Swinging Strike Rate 0.1378 -0.0006 -0.0056 0.0046 0.0006 Ground Ball Rate 0.2368 -0.008 -0.017 0.012 -0.012
Below X-Mov, Y-Mov, Velocity, and Velocity Differential are the coefficients, or the relative weights.
The overall relationship was weighted by number of sliders thrown.
2015 sliders. n=305

That’s a lot of numbers, so let’s break it down in bullet points to make it more understandable.

• The relationship between these variables and those outcomes is not super strong (they explain almost a quarter of the outcomes) but it is significant (p less than .0001).
• Velocity and drop are the most important facets of a slider, whether you’re talking grounders or whiffs.
• It’s harder to predict a good slider for whiffs than a good slider for grounders.
• Velocity differential is not very useful for predicting swinging strikes.
• Velocity differential is very useful for predicting ground ball rates, as useful as velocity itself.
• The smaller the velocity differential, the better the ground-ball rate.

It’s interesting, to consider the slider as a changeup. Obviously Dan Warthen was out in front on this. And it’s not surprising, then, that Jacob deGrom has a below-average swinging strike rate on his hard slider and an above-average ground-ball rate. After all, he’s the patron saint of this “new” slider.

It’s working for him, and for Felix Hernandez. How long till more of the league breaks out the harder sliders themselves? Will it work for all of them? Rubby de la Rosa is wondering, for one.

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.