Let’s Have A Conversation About Spin Rate.

Spin rate is becoming ever more important to baseball analysis now that we have access to more reliable measurement devices. Namely, Trackman. But there are other technologies as well which are being used by high school, college, and minor league teams. Trackman is the big name, though, since it has been adopted by MLB, NPB and KBO along with many colleges and even a few high schools.

Trackman uses Doppler radar to measure the movement of the ball. I want to paint a picture in your mind of what this may look like, in the eyes of the radar. Remember, we’re trying to track the ball here.

Play begins with the pitcher holding the ball. The pitcher is moving, going through his normal motions between pitches. Maybe he stretches his arm and reaches for the sky. He brings the glove in front of his chest, opens it wide, grabs the ball, and looks in towards the catcher. The catcher flashes a few signs, the pitcher shakes off one, nods to the next. He goes to his set position, into the wind up, and there it goes. He threw the ball.

Now imagine you’re the radar. You’re trying to track the ball, but you see wiggles. You see bouncy things. You see chaos. To you, the radar, this is total nonsense. None of this makes sense. It is random noise. All of this walking around, stretching, going into the wind up, etc, all of that is just chaos in your eyes.

Until one moment. The release of the ball. That is when things begin to make sense.

Image source: Evalutation of Doppler Radar ball tracking and its experimental uses. Page 50

This is the world you crave to see: balls effortlessly flying through the air. You gaze upon the ball as it follows a nice, neat, clean trajectory. Then boop. Everything changes in an instant. The ball changes direction. And there it goes. Flying through the air. It is majestic. You live for this stuff. You’re a radar and you’re loving it. Then blang, blerp, bloop. The ball is slowing down. It’s going up and down. Why? Why is this happening?

As it turns out, the player hit a line drive, and it bounced a few times in the outfield.

I want you to hold this image in your head. First, there is chaos. Then there is a smooth line. Then boop. Then a smooth line. Then blan, blerp, bloop. That’s really what the radar is seeing, and if you think about it like that you can make sense of how all of the pieces fit together.

When you go from the initial chaos to a smooth line, that is when the pitcher releases the ball. That smooth line? That’s the pitch. That …boop. That’s contact with the bat. It isn’t as strong as you might imagine. The next smooth line is the batted ball. And then at the end you get the defensive play, whatever it might be.

But wait a second. Let’s zoom in on that pitch part. From back here it looks like a smooth line, but when you zoom in a bit, there is weird stuff above and below it. There are a bunch of weird lines flinging off the ball. It kinda looks like a zebra pattern. What is that?

That’s right, it is the spin rate.

I’m not going to get into the math or the physics of what is going on. If you’re curious about that you can good read this masters thesis about using Doppler Radar in baseball. Let’s keep this on a layman’s level. Those signals coming off the ball, which are picked up and analyzed by the radar, aren’t necessarily a clean rotation. Depending on how you interpret this wave, you might see one rotation, or you might see half a rotation, maybe two, maybe a quarter, or maybe even four. Depending on how you interpret this little tiny wave, you can end up with different numbers of rotations. Since the signal is relatively weak, it might be hard to pick the correct interpretation to use for each individual pitch.

This leads to an issue that you might refer to as “half spin” and “double spin.” A phenomenon where a certain number of pitches thrown by each pitcher may be said to have half or double the spin rate you would expect for a pitch of that type. Some may even have a quarter or quadruple the expected rate. However, the majority of the balls should have a correct reading. In fact, the vast majority of balls should have a correct reading.

To the human eye, it is easy to pick up on these “half spins.” It is very easy to see on a chart.

From here on, I’ll call a half/quarter/double/quadruple spin a “miss.”

This is a segment of a Tableau Viz which you can see here. For now, we’ll focus on the 2015 data. Notice how almost all of the data is set around 2700 RPM, give or take. All of the pitches. Curveball, Fastball, Cutter, Slider. You might refer to this as the main sequence for the pitcher. As velocity goes up so does spin rate. But way off on the left you see these weird, smaller, more scattered pitches.

You might notice they are broken into two smaller groups. One around 1300 rpm, another between 500 and 1000 rpm. These two groups are the half and quart spin groups.

There are a few of the higher spin balls, too. Look closely around 2800 rpm and you can see them for the slider and curveball, and a handful at the top for the fastball. Notice how curveballs and sliders have far more misses than the fastballs.

This is what the half spin and quarter spins look like. To your eye, they are obvious. You couldn’t miss them. To the radar, it is much more subtle.

In terms of analysis, though, clearly this 2015 data will be biased towards lower RPM. I’ve created a GIF that shows average RPM including and excluding the misses. Look at how much the average spin rate changes.

When you include the misses, the average curveball has a spin rate of 2181 rpm. When you exclude them, this jumps up to 2552 rpm. That is an absolutely enormous difference. A curveball spin rate should have a standard deviation below 200 rpm, by my calculations. So a difference of 370 is absolutely bonkers.

Okay, so that is the data from 2015. You may be wondering if the data has gotten better since then? I am happy to tell you it indeed has.

Click the image to see the full sized version.

Each season has fewer misses than the prior. In fact, there are so few misses in 2017 that this issue is practically irrelevant. You can analyze that data without even considering this problem at all.

But, do you really only want to look at 2017 spin data? Or would you rather look at 2016 and 2015 data as well? Because if you’re interested in the 2015 and 2016 data, you’re going to have to do something to address this problem.

If you know the pitcher’s true mean spin rate for each pitch type, you can simply ignore the pitches that might be half or double the true rate. This might be a good solution for many pitchers. But, unfortunately, this will not work for everyone. Take Jeff Samardzija as a difficult to spell example.

Samardzija has three pitches with a velo below 90 mph. I guess we can call them three breaking balls. First, the knucklecurve, or you can call it a curveball if you’d like. Second, the slider. Both of these pitches are pretty standard, and you see them represented along the main sequence of spin rates. Some of them, especially the slider, have a load of pitches that you might consider misses. That’s bad, right? Those are bad measurements!

Say you were going through his pitches and you saw his slider has an average RPM of 2523, but you see there are a load of sliders with an RPM in the 1000-1400 range. A lot of them clustered around 1200 and 1300 rpm. That’s exactly half of the expected result, so you throw them out.

Well, not so fast. Samardzija has a third pitch. The splitter. It has an average velo of 85-87 mph. A bit slower than his 87-88mph slider and a bit faster than his 78-80 mph curveball.  That “half spin” group is in fact a third pitch. MLB is doing a poor job in classifying this pitch, and it is sometimes labeling it a splitter, sometimes a slider, sometimes a cutter, and sometimes a curveball. To further complicate matters, some of these may actually be half spin sliders. What a mess!

To bring this all together, I want you to understand four points.

  1. Quarter, half, double, and quadruple spins exist and cause a tangible issue with the spin rate data.
  2. The 2017 data has been largely cured of this issue, and going forward it probably will not be a problem.
  3. If you want to analyze spin from 2015 and 2016, you need to somehow account for the missed spin rates.
  4. When you account for the spin rate issue don’t forget to take into account potential pitch type misclassification.

Finally, here is the Tableau Viz. There are a number of pitchers included including Chris Sale, Clayton Kershaw, Corey Kluber, Jeff Samardzija, Justin Verlander, and Max Scherzer.


All data used in this piece was lifted from Baseball Savant on the morning of 12/7/2017.

Andrew Perpetua is the creator of CitiFieldHR.com and xStats.org, and plays around with Statcast data for fun. Follow him on Twitter @AndrewPerpetua.

newest oldest most voted
Brad Johnson

Awesome visualization work Andrew. This belongs on the main blog.