On Rotation, Part 2: The Effects of Spin on Pitch Outcomes
On Monday, I looked at how different spin rates for different pitches affect the way those pitches move through the air towards a batter. That post was useful for understanding the relationship between spin and velocity and movement. What it didn’t tell us, however, is too much about what the spin actually does for the pitcher: does more spin make pitches harder or easier to make contact with? Does more spin induce weaker contact? To answer those questions (as well as others), we can look at the actual production from hitters on these pitches. That’s the goal of this post.
The first such stat we’ll consider is contact rate (Contact%), or times made contact (balls in play or foul balls) per swing.
Unlike the graphs showing details about the flight path of the pitch, where the relationship was either constantly increasing or constantly decreasing the whole time for each pitch regardless of the measurement, there seems to be more of a parabolic shape here. (Except for knuckleballs. I don’t even know what’s going on there.) Contact rate is highest when the spin is neither very high nor very low, probably because abnormally fast- or slow-spinning pitches throw the hitter off-guard.
And then there’s swing rate (Swing%), which is just the percentage of times that batters swing at a pitch:
Other than changeups, for which the swing rate increases substantially before decreasing, there’s a very consistent negative relationship between spin rate and swing rate. My guess for why this is the case is hitters have a harder time telling what kind of pitch they’re being thrown when the spin is faster, so they swing less because they’re not totally sure what they’re swinging at.
Let’s also look at swinging-strike rate (Swinging Stike%), which is the intersection of contact and swing rate. It’s the percentage of pitches at which a batter swung and missed, also calculated as (Swing%)*(1-Contact%).
Pretty much what you would expect after looking at the previous two graphs. Throw your fastballs with lots of spin, I guess. All three fastball types (two-seam, four-seam, and cut) see rises in swinging-strike rate as the spin rate reaches the high end.
What happens when the batter makes contact with the ball? Does spin affect what happens then as well? It sure does:
More spin means fewer ground balls (GB%), unless you’re talking about curveballs, in which case it means more. Remember how Collin McHugh was targeted by the Astros because his curveball spun a lot and induced more ground balls? Here’s proof of that in action. And, once again, this makes sense — a topspin-heavy pitch is harder to lift into the air, since its point of contact with the bat (which is usually moving upwards) is moving downwards; and a backspin-heavy pitch is easier to lift, since its point of contact with the bat is moving upwards.
The converse of ground-ball rate, naturally, is fly-ball rate (or, OFFB%):
Not much of interest here. This is essentially the exact inverse of ground-ball rate. But, note, this is only outfield fly balls — we’ll come to pop ups soon. For now, what about line-drive rate (LD%)?
Nope. There’s no real meaningful effect until you get way to the extremes, at which point there are so few pitches that there might as well not be that effect at all. Overall, it looks like there is a very slight downward trend, but nothing very significant. This is not surprising, as line-drive rate is extremely random and unstable.
Pop-up percentage (PU%), on the other hand, gives us something nice-looking:
Other than curveballs, more spin pretty much always means higher likelihood of popups, for the same reason that there are more outfield fly balls with more spin — pop ups are the same thing, essentially, just without traveling as far. Oh, and don’t worry about that rise in the blue curveball curve on the graph after 2,000 RPM. That’s probably random, since there are few curveballs which spin that much. (Click here for a good read about the physics of pop ups.)
Pop ups are good for the pitcher, but the danger with balls in the air is that they can become home runs. Does the spin have any effect on that?
The left side of this graph is extremely jumbled, but the right side shows a huge dip in the rate of home runs per fly ball (HR/FB%) when the spin rate gets very high. I would ignore the curveballs and sliders again, since there aren’t very many breaking balls that spin that fast and that’s just random weirdness. But the two-seam and four-seam fastball dips are very real, as the trend starts before the x values get so high that there aren’t enough data points. This means that faster-spinning fastballs are (a) harder to make contact with, (b) popped up more, and (c) hit for home runs less often. Basically, at least for fastballs, spinnier = better. (Is spinnier a word? I’ll call it a word.)
To put the previous five graphs together: how are batted balls affected by spin rate overall? To answer that question, let’s look at spin rate vs. BABIP and spin rate vs. wOBABIP, a stat I’ve mentioned before and which is just wOBA on batted balls. Since I’m covering multiple years with this, I used the Basic wOBA equation.
Hmm… so maybe the spin rate doesn’t actually affect batted-ball production as much as it does batted-ball type. You can see a decrease in wOBABIP for two-seamers that spin at more than 3,000 RPM and curveballs that spin at over 2,000 RPM, but there aren’t so many data points at those values and that is, once again, probably just random variation. That’s unfortunate, since it looked for a little while like spin rate might give us a clue about which pitchers can keep a low BABIP and beat their peripherals.
Even so, though, we now know more about the effects of spin on pitches. Similarly to my location article from last month, this isn’t anything that will break new ground in baseball analytics — it’s not super applicable — but it’s pretty cool.
Jonah is a baseball analyst and Red Sox fan. He would like it if you followed him on Twitter @japemstein, but can't really do anything about it if you don't.
Warning: This post will be a bit physics-y
While I do not doubt that it is easier to lift a ball with backspin than with topspin, I do not agree with the reason. The physics is actually wrong. To prove it to yourself, drop a spinning basketball onto the floor. Try both spin directions and see which way the ball bounces. It bounces in the opposite direction that the surface of the ball is moving when in contact with the floor. And the reason is very simple. If the surface of the ball moves in one direction, friction acts in the opposite direction, resulting in a bounce in the opposite direction.
The most likely reason it is easier to lift a ball with backspin is that it is usually higher in the strike zone due to the lift that the backspin produces. The opposite is the case with topspin, when has downward movement, ending up lower in the strike zone and making it more difficult to lift.
Please don’t interpret this post as a criticism of an otherwise fine article. I just want to make sure everyone gets the physics right.
I think that makes a lot of sense, but I think that the friction on the surfaces of a baseball and a bat (slicker than a basketball and ground), and the speed at which both are traveling at the time of impact make the physics that you are referencing much less dramatic on the actual outcome of the angle of the hit. I tend to agree with the author’s comment about the angle the ball is moving affects the angle in which it is hit.
Of course, my basketball example was only to illustrate the point. Real experiments with ball-bat collisions have been done that actually prove the point. There are many things that determine the launch angle of a batted ball, including the downward angle of the ball, the upward angle of the bat, the squareness of the collision, and the spin of the ball. My comment only applies to the effect of the spin, not to all the other things. So, if a ball is incident horizontally and the bat is swung horizontally and the collision is perfectly squared up (to keep all these factors fixed), then the ball will go down if the pitch has backspin and up if the ball has topspin. That is an experimental observation, which also happens to be completely in accord with the underlying physics that I described.
Okay then yeah, I see what you mean at that point. Thanks for the explanation.
Ah, much appreciated. You obviously know infinitely more about this than I do, so thank you!
Through this though you have to remember that gravity plays a huge role here. The reason the ball falls without topspin is gravity. You can can negative vertical movement by throwing a changeup off from the side with sidespin to cause horizontal lift moving it horizontally and still getting your vertical movement from gravity. Your curveball is producing negative lift for bonus drop. Backspin holds the ball up.
In my previous comment, I meant to include the following quote from the article;
“a topspin-heavy pitch is harder to lift into the air, since its point of contact with the bat (which is usually moving upwards) is moving downwards; and a backspin-heavy pitch is easier to lift, since its point of contact with the bat is moving upwards.”
It is this statement that I was objecting to.
The knuckleball line in the first graph looks that way because spin rate is guessed based on how much the pitch “moved” and unlike all other pitches, knuckleballs move less the more they spin. So those alleged zero-spin knuckleballs on the left actually spun a lot (for a knuckleball), didn’t move much and got hammered while the ones heading to the right actually had less and less spin, moved more and were harder to hit.
OK, I see your point. You want to argue that the left side of the curve is small movement (I agree) and that it is due to too much spin on the knuckleball (that is one possible reason for small movement–there might be others). But I think we both are in agreement that small movement on a kunckleball results in better contact than large movement, regardless of why there is small movement.
Do you ever think that this could be used by organizations or pitchers that would look at this, especially with the swinging strike% to find the ideal spin rate for a pitch and maybe fiddle with different grips to attain that spin rate? Or is this comment a complete display of my ignorance on how players pitch?
I think it would be used more by organizations trying to target specific players. It’s hard to make adjustments to spin rate without messing with other things in your delivery. But generally, yeah, I’m sure teams are at least generally aware of these trends.
I don’t agree with the comment from TheChaosPath about knuckleballs. The way that I interpret these graphs for knuckleballs is that the spin is simply a surrogate for movement. It is not the actual spin, which is very low. So, less spin on the plot (left side of graph) really means less movement, so the ball gets hammered. Likewise, more spin (right side of graph) really means more movement, so the ball is less likely to be hit well. Perhaps that is what the comment meant to say, but it didn’t seem to come out the right way.
Same reasoning applies to the BABIP and wOBA plots.
We are saying the same thing:
No spin (good knuckler) = High movement = High spin rate guess
Low spin (bad knuckler) = Low movement = Low-moderate spin rate guess
Moderate spin (bad any other pitch) = Moderate movement = Moderate spin rate guess
High spin (good any other pitch) = High movement = High spin rate guess
Yes, we agree completely. Your last post nails it perfectly.
Just one observation: A harder thrown fastball will have more spin (in rpm) than a slower fastball, even if thrown by a robotic arm. Therefore, there is some confounding of variables, in that spin rate is highly correlated, i’m guessing, with velocity at least on fastballs.
I was going to make the same point about velocity as a possible confounding variable, expect that there may not be the strong correlation you suggest. The spin on most of these pitches is highly dependent upon the manner of release, and movement of the hand. Koji Uehara throws 87-88 with insane rotation. In Eduardo Rodriguez’s debut there seemed to be an inverse relationship between his FB velocity and rotation (based on eyeballing the graphs at Brooks rather than looking at the actual numbers, admittedly).
The main point is that any study that just looks at one of a number of factors is going to be approximate, maybe very approximate, as long as the factors are not evenly distributed with respect to one another. For instance, if there are any correlations between rotation and command, that’s going to throw the results very askew.
Maybe this has been answered elsewhere, but if two pitchers with exactly the same arm speed threw fastballs, but one had high spin rates and the other low spin, would the fastballs still have the same velocity?
That is difficult to say without some experimenting. The effect would be meaningful in ballistics but I’m not sure it would be in throwing a baseball. A ball with more spin has more rotational kinetic energy that stabilizes the trajectory and a straighter trajectory should mean a (fractionally) higher effective velocity. At least with gyroscopic spin.
But different kinds of spin would achieve different things. topspin would increase the arc of the trajectory and thus reduce effective velocity, whereas backspin straightens the trajectory and should give higher velocity at all humanly possible spin rates.
However, spherical objects are in theory immune to yaw effects on drag, so the loss of terminal velocity would from not spinning would appear to be very small – and this is the major area of impact for spin in general ballistics. I reckon it could actually be that there is slightly increased drag from the spin as drag increases friction on the high-pressure side of the ball and reduces it on the low-pressure side. But it would come down to how that changes the total size of the wake and I don’t know if there is statistically significant data on it. It must be very minor compared to the overall Magnus effect though.
Cool to see all the graphs in this post and the previous one in this topic. I enjoyed looking at them. Keep it up!
I have a two points though where I would like to chime in:
You look at the swing% v. spin rate graph and draw the conclusion that batters are swinging less at pitches with more spin. Given how hard it is to tell relative the spin rate by looking at a ball (axis of spin is one thing, but rate is very, very hard to do) as compared to judging direction and velocity, i don’t think it is likely that batters are making a judgment on the spin rate. That is to say, I reckon correlation does not imply causation here.
I do think there is a common cause for the drop in swing rate that you did not control for. Going back to your first article on spin rates, it shows that there is a positive correlation between spin rate and velocity on most pitch types, but a negative one for curveballs. That is easily explained by biophysics: let me explain.
At release point, the hand is moving down relative to the trajectory of the ball. With most pitches, the hand is behind the ball and ‘brushing’ down, the ball ‘rolls off’ the fingertips at this point (the contact point shifts). This roll induces the spin. the distance travelled in that bit of rolling is related to the type of motion, not so much to its speed. Thus, the faster you throw a ball, the more quickly the ball rolls off the tips of the fingers and the faster it spins.
Curveballs have the opposite phenomenon because the motion inducing the topspin is somewhat unnatural and thus inefficient. Here part of the motion in the arm is converted not into linear momentum, but rotational momentum as the fingers push ‘over the top’ of the ball. the topspin is the result of the fingers moving over the ball – that is, faster than the ball itself. How much spin depends on the speed difference between the center of the ball and the fingers moving over the top. There is a trade-off: you have to give up ball velocity to gain ball spin.
What I suspect is driving the correlation between spin rate and swing% is common causation by velocity. Higher velocity leads to to more spin. Higher velocity leads to lower swing%.The underlying effect of velocity would come into play in all other correlations v. spin rate, too. This is probably something you’d have to look into if you wanted to derive other causality from the data.
—
As for the knuckleballs, could it be that something went wrong in averaging spin rates? For most types of balls, the spin rate may vary, but it will be within a limited range of angles. Because knuckleballs are meant to roughly cancel out spin, there will still be some spin but spin should be distributed near-randomly over all angles. If you take the scalar sum in averaging, all the spins add up, but if you take the vector sum in averaging, all the spin would cancel out.
I’m not sure what to make of it or what correlation to predict. A knuckleball with more spin would be more surprising (and effective) if the spin is distributed randomly, but if the spin is not distributed randomly, then less spin would be better as less spin leads to a more unstable (and unpredictable) trajectory (rotation stabilizes trajectory of travel because of complex newtonian physics) which is what knuckleballs are about.