# Spin Rates, Swinging Strikes, and an xSwStrk Stat.

This is the first year we’ve had access to the new ‘Release Spin Rate’ stat, which can be found hiding in a little nook of Baseball Savant. This spin rate, as I understand it, is measured using Doppler Radar at the moment the ball leaves the pitcher’s hand. I’m not exactly sure how the system defines this release point, perhaps when the ball begins to slow down. No matter the case, this new ‘Release Spin Rate’ stat appears to have some potential as a new way of evaluating pitcher performance, since we all assume there must be some correlation between spin rate and success rate as a pitcher.

Lets get some Physics out of the way.

I want to preface this by saying this isn’t meant to be a physics lesson, I’m intentionally oversimplifying everything. I just need to cover some basics before I can move on. I’ll post some links at the bottom if you’re interested in more information.

When a ball is spinning, half of the ball is moving in one direction, and the other is moving in the opposite direction. For example, as the earth rotates, half of it goes into sunlight and the other half goes into darkness. If you move this spinning ball, one part of the ball will be both rotating and moving in the same direction, while the other side of the ball is rotating opposite to the direction of movement. The part of the ball that rotates towards the direction of its movement fights against the air that is trying to brush past it, and it builds a little high pressure region in the air as it moves. You can think of it as the ball pushing against the air, and as you know from the third law of motion, every force has an equal and opposite force. When the ball pushes against the air, the air pushes against the ball. It pushes from this high pressure region, pushing the ball away from the high pressure. This force is called the Magnus Force.

So, for example, if you release a four seam fastball, the ball will be spinning with back spin. That means the bottom of the ball is spinning towards the direction of movement, making high pressure under the ball. This high pressure pushes the ball up. As a result, a four seamer will to rise. Gravity is much stronger than this Magnus Force, so the ball never actually stops dropping, but it does end up dropping less than you would have observed had you thrown a ball without spin.

This force is proportional to quickly the ball is spinning, and the force only works when the spin is perpendicular to the direction of movement. So, if you spin a ball like a bullet, where the axis of rotation is parallel to movement, then the spin will not move the ball (which is why you want to spin a bullet in this manner!). If you spin the ball on an angle, then only a portion of the spin, given by the cosine of the angle, will count towards movement of the ball. In essence, there are three types of spin on a ball: Front/Back spin, Side spin, Unhelpful spin.

As I noted before, back spin makes the ball rise. On the flip side, front spin makes the ball dive downward. Side spin makes the ball move left or right. The third category, the unhelpful spin, does not affect the movement of the baseball.

You may be thinking, wait a second, haven’t we had a “Spin Rate” stat for the past few years using Pitchfx? Well, yes and no. We did have some knowledge of the spin, kind of. It was never actually measured but rather estimated using the known movement of the pitch. For example, if a pitch weren’t spinning it would have gone here, but since it went 3 inches to the left, it must have been spinning. In order to move 3 inches, it must be spinning at least this much. In other words, we could estimate the total useful spin, the front/back and side spin, but we never knew exactly how much wasted spin a pitcher may have, and it was all an estimate. The ‘Release Spin Rate’ stat is, from what I can tell, directly measured using Doppler Radar. Statcast also has a ‘Spin Tilt’ stat, which tells us which angle the ball is spinning as it leaves the hand. I’m not fully acquainted with exactly how this tilt is measured, and I’m not using it here.

Visualizing how Spin Rates Affect Swing and Miss

Okay, with the basics out of the way, and, again, I’ll have links at the bottom for more thorough explanations for all of the above, lets get into how SwStrk% works into all of this.

I’ve split the strike zone into 4 inch squares, a 16 x 16 grid, ranging from -2.5 to 2.5 feet in the x direction, and 0 to 5 feet in the z (vertical) direction. I’ve also grouped spin rate by 100 rpm increments, and pitch velocity in 2 mph increments. I’ve gone through and calculated the SwStrk% for spin rate vs x coordinate, z coordinate vs spin rate, and pitch velocity vs spin rate. You can (hopefully) see them embedded in the charts below. Feel free to check out the charts in all their glory, you can hover over elements to see more information, use a few simple filters, etc. In these charts, anything Green represents above average, anything Blue is below average, and everything Grey is roughly average. Average is being defined as ~10% Swinging Strikes.

These charts look significantly better in their full screen version.

A few quick notes. Some pitches do not come complete with Release Spin Rate data, and a few are missing pitch velocity as well. Only pitches that contain both of these stats are included in this analysis. You will also see a few obvious outliers in these charts, areas with 100% swinging strike rate, for example. The data ranges being used are pretty large, especially when pitch velocity goes above 100mph or spin goes above 3200 or so, so the edges can have these problems. I have opted to keep these, I could have just as easily clicked ‘exclude’ and hidden them away. It is a stylistic choice, I prefer to include all of the data without messing with it too much. Take the high spin and velocity rates with a grain of salt.

X Direction

Looking at the Spin vs X chart, note that the area between -.8 and + .8 are, roughly speaking, the width of home plate. You can see over this area there is a little valley of low swinging strike rate between 1900 and 2300 rpm. This matches up pretty well with the spin rate of your average fastballs. You definitely do not want to throw straight pitches in the zone. This isn’t ground breaking stuff here, but it is interesting to see it broken down in this manner, especially seeing those extremely high swinging strike rates up towards the 3000 rpm region.

Z Direction

Keep in mind the strike zone in this chart is roughly between 1.67 and 3.33. You can see a cross shaped area of low swing and miss in this chart. On the spin side of things, you see that same gap between 1900 and 2300 rpm, but this time, on the z axis side, and remember this is height above the ground in feet, there is a big gap in the region that roughly approximates the strike zone. The bottom 4 inches or so of the strike zone has pretty good SwStrk%, but for the most part the zone is barren of swinging strikes. It is interesting to see the little island of high swing and miss at the top of the zone, between 3.3 and 3.6 feet, hovering around 2200-2600 rpm.

The Velocity Chart

Yesterday Jeff Zimmerman posted charts similar to this, except limited to fastballs and curveballs. The chart I made, though, is for all pitch types. To me, this chart looks a bit like a continent, and if you look around you can probably make out a few regions, (countries?) based upon their velocity and spin rates. You see a breaking balls in the top right corner, ranging from about 1900 rpm on up, and topping out around 86 mph. You see as the velocity and spin rate of the pitch goes up, as does the effectiveness of the breaking ball.

In the middle, towards the bottom, you see the Bay of Cheese, that blue region between around 1700 and 2500 rpm and 86 and 100 mph. For the most part, this region is very blue, meaning very few swinging strikes, but the high ends of both fastball spin and velocity you do reach land on the Cape of Heat. Those pitches can have very solid swinging strike rates, up to 12% and above.

North of the Bay of Cheese you see the Land of Dead Fish, home of the change-up. Centered around 1700 rpm and 83 mph, this area has pretty good swinging strike rates. As you travel south, a little higher in velocity but lower spin, down towards 1500 rpm and 86 mph, you bump into the splitters and sinking fastballs. They rub shoulders with the deep blue ocean of low swinging strike rates, so you probably want to keep the velocity down if you want to miss bats in this area.

Between these regions I’ve noted you can find other areas, sliders and cutters mark the changing landscape between fastballs and true breaking balls while various types of sinkers and off speed pitches litter the left half of the chart.

Bringing It All Together

Those three charts are pretty cool, and honestly I could spend a lot of time playing with them. I had to pry myself away to write this. By the way, if you check out these charts later on today or over the weekend, I may have added something. I’d love to add the number of swinging strikes and total pitches thrown to each element, maybe the number of each pitch type as well. But first, I want to show you what I’ve done in combining these charts to create an xSwStrk% stat.

I have taken each of these dimensions, the x and z coordinates, velocity, and spin rate and combined them with the x and z movement to classify every pitch thrown over the past two years. For each of these classifications I find the average SwStrk% for the bucket, and assume every pitch thrown similarly will have that average Swinging Strike rate. About 9% of the pitches thrown must be dropped due to missing data points, so the SwStrk I’m calculating reflects only the pitches I can work with.

When I calculate the xSwStrk for batters, xSwStrk% has a .81 correlation with SwStrk% of the total pitches and .82 with the pitches usable for the calculation. I’m going to attach an excel spreadsheet below that contains the Actual Pitch Count, Usable Pitch Count, SwStrk%, SwStrk% of the usable pitches, and xSwStrk% for each pitcher in the past two years on one sheet, and just this season on the second sheet.

Looking at 2016 alone, the correlation drops to .78. This could be, in part, due to an increased number of missing data points. The missing data accounts for 11.6% of the pitches in 2016 versus 8.92 in 2015 and 2016 combined. This season also has a higher overall swinging strike rate, and the data from last year, since it is weighted equally, may be bringing down the totals a little bit. Either way, these are pretty solid correlations, so the system seems to be doing something right.

I’d like to address the elephant in the room here. There are a lot of pitches that are classified either by themselves or with very small numbers of other pitches. About half of the pitches are combined with at least 10 other pitches, giving pretty decent expected results. But then a few hundred thousand are in groups of 2 or fewer. This could be a major flaw, it may not be. I’m not sure. This problem made me hesitate writing about this work in the first place, but after thinking about it for a while I figured I may as well throw it all out there.

This data is largely new to me.  I could quote pitchers who have over and under performed the xSwStrk%, but I’m honestly not really sure what it means to do so, and I’m putting this out there hoping to get a little feedback on this.  It could be that a difference in xSwStrk and SwStrk represents deception, one of the most notoriously difficult to quantify skills in the game.  That could be a pretty good thing, right?  Perhaps the pitfalls of the stat, especially the way pitches are bucketed, washes everything out and there isn’t much usable information here.  Maybe this is a way of estimating the true swing and miss skills for a pitcher. Honestly, I’m really not sure, and if nothing else I hope you enjoy those charts I made.

Here is some extra information for those interested.  The videos by Veritasium are very interesting and do a good job reaching out to those who are interested in the information but not as keen about digging through the actual math and science.  The article about Arbitrarily rotating spheres is a bit more technical, and the Baseball Prospectus article is somewhere in between.

The motion of an Arbitrarily rotating spherical projectile and its application to ball games.

Measuring pitching with Trackman. from Baseball Prospectus

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

2 Comments
Inline Feedbacks
View all comments
John
7 years ago

Did you happen to run the correlations for xSwStrk and SwStrk in 2015 against SwStrk in 2016?