# The Lurking Influence of Batted Ball Spin

If I may, I think the uncertainty regarding this season’s offensive environment has made us a bit paranoid. Are hitters lagging behind pitchers due to an irregular spring training? Is the ball not traveling like it once did because it’s been replaced yet again, or is the mass introduction of humidors to blame? Or worse, has MLB introduced multiple balls into the game, some of which are being used in certain games to boost action or influence outcomes?

That last theory has been floating around my Twitter feed for a while now. I’m not going to discuss whether it’s true, but I brought it up because supporters of the multiple ball theory will often compare two batted balls with near-identical exit velocities and launch angles. One ends up traveling more than the other, despite all the indications that it should not. Aha! *Something* must be up.

In response, a lot of people have suggested batted ball spin as an explanation. Maybe one ball came off the bat with backspin and the other came off with topspin, which would drag the ball down as it traveled through the air. Unfortunately, since data on batted ball spin isn’t available on Baseball Savant, this might seem like a dead end. Don’t worry, though: I had some leftover Trackman data on 2021 NCAA Division I baseball games from a piece that Eric Longenhagen and I collaborated on during last year’s Draft Week, and they contain mostly complete readings on the spin of a batted ball. Could we use collegiate baseball to learn about the odds and ends of batted ball spin, and what it tells us about hitting?

You bet. But first, I built a quick and dirty generalized additive model to estimate how far a ball should travel using just exit velocity and launch angle; let’s call it xDistance. If there’s a positive (or negative) difference between a ball’s actual and expected distance, that could suggest the influence of batted ball spin. Because displaying the more than 30,000 balls in the dataset is a headache, I settled on a subset that seemed interesting: fly balls that barely qualify as being “hard-hit” — balls that would end up a home run with a little extra distance but are a warning track fly out otherwise. In such cases, spin can often be the deciding factor. But it’s very much unclear which outcome will show up, as the graph below shows:

Since Trackman data doesn’t provide batted ball spin axis, it’s impossible to distinguish between backspin and topspin. Only raw spin rate is available, which forms the x-axis. Still, there are a ton of takeaways from this single graph. You can see from the wide range of actual minus expected distance differentials that the same amount of spin can have a drastically different impact on the ball’s flight. Together with wind speed, direction, and the weather that day, some 2,000 revolutions per minute are capable of adding upwards of 40 feet — or subtracting it.

Also, while raw spin rate doesn’t show a correlation with actual minus expected distance in small doses, it’s past the 3,000 rpm threshold that a negative relationship begins to form. Why is that the case? Too much topspin isn’t great news for a ball attempting to travel far, but so is too much backspin; that just keeps the ball hanging in the air for a long time, allowing fielders to track it down. The parabolic path of a batted ball isn’t solely determined by launch angle; spin matters, too. Save for that one eyesore of an outlier (which is possibly a misread), the vast majority of balls in the 4,000-plus rpm range are in the red when it comes to actual minus expected distance. They’re less common, but they do exist, and it isn’t hard to imagine that two seemingly identical batted balls can take on divergent flight paths.

Let’s consider individual hitters as well. As this FanGraphs community post demonstrated years ago, a hitter’s fly ball rate is strongly correlated to the amount of distance he *loses* on his batted balls. It sounds counterintuitive, but as tennis shows, it’s actually an uppercut swing that provides the interaction necessary to generate topspin.

One aspect that article doesn’t touch on, however, is the day-to-day inconsistency in the batted ball spin that a hitter creates. This is true of even the most extreme hitters. Take Tre Morgan for example: a first baseman at LSU who has the sixth-lowest fly ball rate (27.7%) and the second-most distance gained on his batted balls (11.8 feet) among his Division I peers. Here’s a simple graph showing Morgan’s actual minus expected distance in each game he played in April:

Owing to Morgan’s fondness for grounders, most of the bars here are pointing north. But there’s a lot of variation as well, with certain games ranging from as little as four feet of added distance to a whopping 40 in the last one of April. Other factors also interfered with how these balls traveled, but this still goes to show that though individual hitters do have tendencies, they appear in the aggregate.

Next, let’s take a look at Troy University third baseman Drew Frederic, who unlike Morgan has a high fly ball rate (49.4%) and the second-most distance *lost* on his batted balls (-13.3 feet):

Frederic’s bar graph looks like a modern art masterpiece more than anything — he didn’t play much in April — but I stuck with the month for consistency’s sake. Again, the bars vary in direction and length, with some games more at the whims of batted ball spin and miscellaneous elements than others. We can assume Frederic’s (and Morgan’s) swing didn’t change all that much throughout the month, meaning there’s a limit to how much hitters can control. With enough practice, one could certainly learn to hit the ball further. But keeping the spin imparted to it consistent? That’s a whole ‘nother problem, it seems.

What about someone not at either far end of the launch angle spectrum? Meet Gray Betts, the Liberty University catcher who’s about smack-dab average in fly ball rate (39.7%) and distance added (1.4 feet). Here’s the same graph as the ones before using his data:

Huh, that’s weird. It looks like Betts’ batted balls are a lot wilder than Morgan or Frederic’s. Check out the mid-month swing from negative to positive and then to -20 feet of distance again. Are average hitters by fly ball rate more at the whims of batted ball spin?

To answer that question, I split the hitters I had into four different groups according to their respective fly ball rates, then calculated the standard deviation in actual minus expected distance of each group. Since standard deviation measures how close the values are to the mean of the set, a larger one should indicate back-and-forth jumps between backspin and topspin.

The results suggest “no” as an answer. There’s hardly a difference between the hitters with standout batted ball tendencies and the ones with normal tendencies. The distribution of individual hitters’ actual minus expected distance is fairly normal, too. In other words, I didn’t find any evidence of “batted ball spin consistency” being an actual skill. By the way, Betts ranks 176th in standard deviation among the 176 hitters in my sample — dead last. It turns out that his four-game mid-April parkour doesn’t happen in any other month, or even week for that matter. This is yet another reason to grade the eye test using data; our eyes are terrible!

It’s time to list the caveats, of which there are a few. For one, Trackman is nowhere near as accurate as Hawk-Eye, which is Baseball Savant’s source. Each school has its own setup, resulting in various discrepancies. That also resulted in a couple hundred rows discarded due to missing values or their apparent weirdness, such as a launch angle that’s an entire standard deviation above even the normal outliers. Also, I found a curious lack of batted balls below zero degrees, which likely affected the calculations of hitters’ fly ball rates. Most importantly, the inability to distinguish between backspin, sidespin, and topspin made it difficult to figure out certain relationships, such as the per-rpm benefit of backspin or the possible trade-off between different types of spin.

Lastly, it’s good to remind ourselves that batted ball spin isn’t the only variable worth considering. Not only are there the aforementioned weather-related effects, but there’s also batted ball direction: pulled, straightaway, or opposite. How is that relevant? We already know straightaway or opposite fly balls have a harder time adding distance than their pulled counterparts. But if the baseball this year really does have a higher drag coefficient than last year, it would, in theory, add an extra degree of difficulty. That also means the same combo of exit velocity and launch angle is no longer as reliable as before. We’ve been used to a drag-less baseball for a while now; it may be time to adjust our expectations.

But overall, the point is this: If you’re wondering why two balls with similar exit velocities and launch angles don’t travel the same distance, batted ball spin may have a lot do with it. There really are cases where the ball loses 40, 50, or even 60 feet of expected distance. And as we explored earlier, other than a general penchant for presumed topspin or backspin, hitters don’t seem to have much control over it on a daily basis. While the numbers smooth out in the end, some days a hitters will impart too little spin, and on other days he’ll impart too much of it. With all this in mind, it’s plausible a dramatic gap in distance appears in the same game between different hitters. It’s not on any Savant leaderboard (yet), but batted ball spin is most certainly a thing.

Justin is an undergraduate student at Washington University in St. Louis studying statistics and writing.

Can the physics knowers help me out as to what causes fly balls to have different amounts or types of spin, especially if you’re talking about those hit at equal launch angles? I’m guessing it has to do with the angle the pitch came in, the spin of the pitch, and the angle of contact (pull/oppo) but I’m purely speculating and I suspect someone(s) here knows more than me

All of those, I think, but also (maybe even mostly) the swing plane. As Justin noted, an uppercut swing tends to generate more topspin, while a flatter swing plane would give a bit more backspin. So that’s largely driven by the batter’s swing mechanics, but also affected by pitch location and the path the bat has to take to make contact.

I imagine somebody in a MLB team’s analytics department has done a deep dive on this and modeled batted ball spin and expected batted ball results for various pitch types, pitch locations, and swing mechanics. From where we sit, all the variation in batted ball spin and lost/gained distance mostly looks like noise because we can’t see all the correlations, but I wouldn’t be shocked if there’s actionable information in there like “Player X, don’t be afraid to swing at a curveball at the bottom of the strike zone, because if you make decent contact it’s going to carry a ton.”

Coefficient of friction at ball/bat interface is going to matter too.

It’s often helpful to strip things down to the simplest situation (fewest variables) and then build it back up again. Imagine a 2-dimensional disk, with no spin, hanging in space. Imagine a second disk striking it; this second disk has a vector of motion (visualized as an arrow that passes through the center of the disk, indicating the direction and magnitude of that motion). If the motion vector of the striking disk passes through the center of the struck disk, no spin is imparted; if the vector passes below the center, some backspin is imparted, whereas if the vector passes above the center, topspin is imparted. The amount of spin will be proportional to how off-center (or “oblique”) the collision is. And, separate from that, the initial direction the struck disk takes after the collision will also be affected by the vector of the striking disk and how off-center or oblique the collision was. If you’ve ever played air hockey, you already have a sense of how this works.

Now start adding back in the variables: if the struck disk is already spinning, the offset nature of the collision will add or subtract to that spin*; the collision is not instantaneous, and the disk deforms during it, creating friction that will have more or less effect on imparting spin; and of course all of this is happening in three dimensions with collision involving a tapering cylinder and a rough sphere that is already spinning around multiple axes. It’s complicated — though if you’re just thinking about spin that imparts more or less carry on the ball, and you’re not an outfielder trying to track a ball slicing or hooking across the field, you can ignore a lot of that.

*However, Dr. Alan Nathan has found, “For a given angle of incidence, the scattered spin is nearly independent of the incident spin;” — in other words, the spin the pitcher put on the ball doesn’t matter and essentially all of the spin after the bat-to-ball collision was the result of the collision.

If you want to see an illustration of this with a real non-spinning baseball hitting a bat and having spin imparted as a result of the off-center (“oblique”) collision, have a look at the high-speed video Dr. Nathan posted on his page (and then, if you want to dive deeper, read the paper and slides he links on that page). Note that this work was a decade ago and there has been a lot of analytical progress (by him and others, including some teams) since then.

This is a really helpful way to imagine the collision of bat and ball. Thanks!

The last thing you need are the physics powers. It matters how you hit the ball and it can happen on any pitch in any location. Some guys have a great ability to backspin and some don’t.