This season is the third since the implementation of a spate of significant rule changes across the majors. Along with a pitch clock and limits on defensive positioning, a limit on disengagements (read: pickoff throws plus idle standing around) combined with slightly larger bases gave runners a collective green light. With fewer throws to first, bigger targets to slide into, and more predictable pitcher deliveries thanks to the clock, stealing a base got much easier overnight. In 2022, the last year of the old rules, the majors saw 2,486 steals across the entire season. In 2024, that number surged to 3,617 steals. Even better from an offensive perspective, the stolen base success rate jumped from 75.4% to 79% over that span.

The first year of the new rules was all about experimentation. Some players ran wild – Ronald Acuña Jr. more or less took off every time he could. Meanwhile, the Giants stole just 57 bases as a team, fewer thefts than the previous year, when those steal-boosting rules weren’t yet in effect. None of that seems particularly surprising to me; when new rules of this import are added to the game, every team will scramble to figure out how to change their own behavior to benefit. There were a ton of moving parts, and many teams took a simple approach: keep stealing more and more until it starts to fail.

The 2024 season was the year of the defensive reaction. Teams attempted 209 more steals in 2024 than they did in 2023, but only succeeded on 114 of those extra steals. The aggregate effect was a lower success rate on marginally more attempts. Catcher pop times improved, pitchers threw over more often, and defenses were more attentive to baserunners in general. That brings us to 2025, and in the early going, it looks like the baserunners are continuing to push the envelope:

Read the rest of this entry »

To every thing there is a season, and a time to every purpose under the heaven:
A time to be born, and a time to die; a time to plant, and a time to pluck up that which is planted;
A time to kill, and a time to heal; a time to break down, and a time to build up;
A time to weep, and a time to laugh; a time to mourn, and a time to dance;
A time to cast away stones, and a time to gather stones together; a time to embrace, and a time to refrain from embracing;
A time to get, and a time to lose; a time to keep, and a time to cast away;
A time to rend, and a time to sew; a time to keep silence, and a time to speak;
A time to love, and a time to hate; a time of war, and a time of peace.

– Ecclesiastes 3:1-8

On Thursday, Toronto Blue Jays outfielder Nathan Lukes welcomed a baby named Jett into the world. That same day, teammate Daulton Varsho was expecting to welcome his own baby. When I read the news, I did what anyone would do. I thought, “How wonderful for the Blue Jays,” and then I asked the internet to do some math for me.

Well that’s fun. Nine months before Lukes and Varsho became fathers, it was July 16, 2024. That date may ring a bell, because it was also the date of the All-Star Game. Lukes wasn’t in the majors at the time, but clearly, both players had very productive All-Star breaks. I decided it was time for a full investigation. Do baseball players make all their babies during the All-Star break? Read the rest of this entry »

Last year, David Appelman and I set about injuring a ton of players. Wait, that doesn’t sound right. Let’s try this again. Last year, David Appelman and I developed a method to use our depth charts projections to simulate how much injuries to the league’s top players might affect each of the teams in baseball. Today, we’re updating that article for the 2025 season. I’ll also present some research I’ve done into how these injury-aware depth charts compare to actual historical seasons.

First, a review of the methodology is in order. If you don’t need an update, or if you simply want to get right to the data, you can skip ahead; the results section is clearly labeled below. We decided to simulate depth by first removing the top X players from a team’s depth chart and then reallocating playing time to fill in the missing plate appearances or innings pitched. We then created a number of rules to make sure that these new depth charts were generated in a reasonable way, at least to the greatest extent possible.

Let’s use the 2025 Phillies as an example. As of the time of our run on April 7, we projected the Phillies for a .545 winning percentage against league average opposition. That projection comes from allocating playing time to each Phillies player according to our depth charts, using blended projections from ZiPS and Steamer to estimate the talent level of those players, and then plugging those projections into the BaseRuns formula to estimate runs scored and runs allowed. But those projections have an obvious weakness: they’re static. Read the rest of this entry »

I thought I had it sorted out; it took one batted ball to convince me otherwise. It was the bottom of the seventh inning during the Rays’ first game in their new George M. Steinbrenner (GMS) Field digs. Jonathan Aranda worked his way into a 2-0 count against Rockies reliever Tyler Kinley. With runners on second and third, one out, and the Rays down two runs, Kinley hung a slider. Aranda uncorked his A swing, launching the ball deep to right field. Off the bat, I thought it looked way gone. It didn’t even go 300 feet:

Read the rest of this entry »

As I write this, the winter free agency period has essentially drawn to a close. Out of the top 50 free agents I highlighted before the offseason began, 48 have found homes — sorry, David Robertson and Kyle Gibson. Per RosterResource, only five free agents – including the two holdovers from the top 50 – accrued 1 WAR or more in 2024 and haven’t yet signed new deals. In other words, all the signing that is going to happen basically has, so it’s time to look back and see how you and I did at predicting the deals players would sign.

I like to evaluate my own predictions in service of making better ones in the future, dividing them up into a few categories. First, I break signings down by position, because the market for relievers and second basemen is different. Second, I look at both average annual value and total guarantee. There’s no set ratio for how to relate those two, so looking at each independently seems best to me. Finally, I look at both the individual predictions (how close to the actual contract that a player signed my predictions came), as well as the overall trend (how my aggregate predictions for each position group did compared to the total amount they received).

This year, I made all of that back-checking more rigorous. I put all of my predictions, as well as every crowdsourced one, into a giant spreadsheet. I noted all the contracts that were signed, made adjustments for deferrals, and ignored non-guaranteed money. I compared each actual contract to our predictions. I also gathered some of the best non-FanGraphs predictions I could find, looking to outlets like ESPN, The Athletic, and MLB Trade Rumors. Below, you’ll find how both the crowd (you) and I did, as well as the best non-FanGraphs entrant in each category. Read the rest of this entry »

Reader, I gasped.

Sorry, I’ve just always wanted to write that. I’ll tell you later on why I gasped. Let’s start here. Last Tuesday, while making his Grapefruit League debut with the Blue Jays, Max Scherzer challenged a pitch. Then he challenged the challenge system. Scherzer’s start against the Cardinals marked his first experience with the automated ball-strike system, which is being rolled out in some spring training facilities this year, continuing its inexorable, years-long creep toward implementation in regular season games. Scherzer, for one, does not welcome our new robot overlords.

On his 11th pitch, Scherzer fired a 1-0 fastball to Lars Nootbaar, just clipping the outside corner, at least according to home plate umpire Roberto Ortiz, an organic life form who uses an inefficient pair of weird, goo-filled orbs to assess pitches. Nootbaar – who, we should note, played some rehab games in Triple-A last season, and so was at least somewhat familiar with the challenge system – immediately patted his head. That’s the official way to request a challenge (though I would strongly encourage the league to require the player to rub their stomach with their other hand too). Scherzer, never a fan of waiting around when there’s pitching to do, canted his head from side to side like a racehorse in the starting gate. The machines spoke: The pitch was 2.3 inches off the plate, or as the humanoid Buck Martinez put it, “way outside.”

Just like Scherzer, this was my first experience with the challenge system, and I found the graphic adorable. That’s the point, I guess: implement an all-seeing eye that judges everything and everyone with detached, ruthless precision, then soften it with a lovable cartoon face. Scherzer recovered to strike out Nootbaar, then made his own challenge in the second inning. The right-hander, who tracks his pitches using the same goo-based technology as Ortiz, didn’t agree that he’d missed low with a 1-0 curveball to JJ Wetherholt, and he pounded the top of his cap like a bongo drum.

Reader, that’s when I gasped. Then I laughed. I gasp-laughed. According to the delightful ABS graphic, the pitch was hilariously low. It was nowhere near the strike zone. This pitch was in the Cactus League. It was so far away that Social Distortion wrote a song about it called “So Far Away.”

In fairness, we should acknowledge a few things. First, one of the consensuses that emerged during last year’s test of the challenge system was that team’s should disempower the pitcher from making them. Catchers are right there, and they have a much better sense of the actual location of the pitch. Second, Scherzer indicated after the game that the challenge was more an experiment than an expression of his certainty that the pitch had clipped the zone. “That was a rare occurrence for me, with a curveball down, to actually see if that’s actually a strike or not,” he said. You’re allowed to take that notion with a grain of salt. Part of me believes Scherzer, but, uh, he was bopping himself on the head with a lot of conviction.

We should also note that the steep shape of a curveball makes it hard for the pitcher in particular to judge the exact spot where it crosses the plate. The really interesting thing is that curveballs are actually relatively easy for umpires to judge. That’s not necessarily intuitive. Curveballs approach the plate at such a steep angle that they hit the catcher’s glove (or the dirt) far lower than they cross the plate, which might fool the umpire into thinking a pitch was lower than it was. And curveballs that come in at the very top of the zone leave the pitcher’s hand so high and possess such a loopy shape that they also might be hard to recognize as strikes. Where a fastball or cutter pushes straight through the zone cleanly, a breaking ball slices through it at an oblique angle, and it just seems logical that the more of the zone a pitch catches, the more likely it is that it will be recognized for doing so. But apparently that’s wrong. I broke down the 2024 stats for curveballs and fastballs (four-seamers, sinkers, and cutters) in three areas of the zone: the heart, the top and bottom of the shadow zone down the middle, and the top and bottom thirds of the zone down the middle. I’ve highlighted those areas in pink.

In all three areas, Curveballs had higher strike rates than fastballs. On pitches over the heart of the plate, it was a matter of a few tenths of a percent, but in the middle graph, curveballs were ahead 83% to 80%, and on the right it was 83% to 81%. Maybe it’s just that curveballs are easier to judge because they’re slower, but umpires are better at recognizing when they’re strikes, so in that sense Scherzer picked a bad pitch to challenge. However, much like Scherzer’s curveball, we’re drifting away from our main objective here. We’re focusing on how far the pitch was from the zone, and just to reiterate: It was far.

However, you might notice something about that graphic: There’s no distance measurement. When Nootbaar challenged in the first inning and earned his Nootbaal, the graphic zoomed way in to show us the exact size of the miss down to a tenth of an inch.

When Scherzer challenged, no measurement popped up, and I suspect that I know why. I think this is a deliberate decision made to avoid embarrassing a player who challenges a pitch that’s not particularly close. Nearly all challenges that end up as balls will show the miss distance. But if the pitch doesn’t even touch the shadow zone – that is, if it’s not even within one baseball-width of the strike zone – the graphic leaves off the exact distance so as to avoid blowing up the pitcher’s spot. Max Scherzer, trailblazer that he is, has showed us that although robots don’t feel, they can still be programmed to blush.

Don’t worry. We’re still going to blow up Scherzer’s spot. Because of all the cool graphics, it’s still really easy to get an exact measurement for pitches that land in the Zone of Embarrassment. We know the measurements of just about everything else on the screen. We know the strike zone is exactly 17 inches wide and the ball is approximately 2.9 inches wide, and through the magic of Statcast, we know that because Wetherholt is 5-foot-10, his strike zone is roughly 18.55 inches tall. I threw a screengrab into Photoshop, measured each of those constants, then used the ratio of pixels to inches to calculate the distance. The ball was 3.98 inches from the strike zone. It missed the shadow zone by more than an inch. It crossed the plate just over a foot off the ground.

That looks pretty damning, but allow me to blow your mind for a moment. If we’re being fair to Scherzer, we need to acknowledge that the pitch was actually much closer to the rulebook strike zone than Statcast makes it look. Let’s think about it under the rules of the current, non-computerized strike zone. Keep in mind that this was a curveball breaking downward. Now let’s look at the way that the Hawkeye cameras measure a pitch, courtesy of an MLB.com explainer by Anthony Castrovince.

Keep your eye on the diagram on the right. Statcast’s strike zone is two-dimensional, and it’s measured from the very center of home plate. That’s a perfectly reasonable way to design an ABS system – an earlier version was 3-D, so it seems safe to assume that this 2-D version is, for some tangible reason, an improvement upon it – but it’s not the way the strike zone has worked for the entirety of baseball history, including right now. The rulebook definition starts like this: “The STRIKE ZONE is that area over home plate…” and that’s really all we need to know. The strike zone is three-dimensional. It’s seven-sided, a pentagonal prism, and the ball just needs to clip any part of it in order to be a strike.

At The Athletic, Jayson Stark had the good fortune to be present in the clubhouse after the game, when Scherzer found out that the robo-zone didn’t match the rulebook zone: “Wait, I thought it was the whole plate,” he said. “So now we have to redefine what the strike zone is? You said it was a 3-D zone. Now we’ve got a 2-D zone? Hasn’t it always been a 3-D zone?” The answer to that question is yes. It has always been a 3-D zone and it still is, but now there’s also a 2-D zone. There are two strike zones. We’ll dig into the philosophical implications of this dichotomy later, but for now, that’s how umpires are judging pitches, so why don’t we try measuring things that way?

Let’s start with how curveballs work. Their path gets steeper and steeper as they approach the plate. There are plenty of reasons for this. Curveballs actually leave the pitcher’s hand traveling slightly upward; the classic way to recognize a curveball is seeing it jump up out of the pitcher’s hand. The magnus force created by the ball’s topspin pulls it downward, and that force compounds upon itself over the length of the pitch. Here’s where it starts. The baseball is traveling horizontally, and the topspin interacts with the air to start pulling it down.

Now it’s traveling at a steeper angle, but guess what? The topspin is still pulling it downward, so its angle is going to keep getting steeper and steeper as it goes.

Moreover, gravity amplifies this effect for a pitch that’s always breaking downward. Air resistance slows the pitch down as it nears the plate, but gravity is pulling the pitch downward at a constant rate. So say it takes a tenth of a second for the ball to travel the first 10 feet toward home plate, and in that time, gravity pulls it down five centimeters. By the time it reaches the plate, it’s going slower, so over the last tenth of a second, it only travels eight feet, but gravity is still pulling it down five centimeters. All the numbers in this example were completely made up, but you get the point; the ratio of downward movement to horizontal movement is increasing. A curveball’s approach angle keeps getting steeper. You can see it in Statcast’s 3-D pitch visualizations.

These are two actual Scherzer curveballs from last year. We’re going to focus on the bottom one, which came in a bit below the plate. The red line shows a straight line between the position of the ball when it crosses home plate and the position when it’s 50 feet away.

Now, let’s zoom in and look at the path of the pitch over the last few feet of its journey. As you can see, our new purple line is significantly steeper.

None of this should be particularly surprising if you’re familiar with Alex Chamberlain’s primer on vertical approach angle, but the point is that curveballs, with their sharp downward movement compounded by gravity, are the steepest pitches of all. According to Alex’s pitch leaderboard, Scherzer’s curveball averaged a vertical approach angle of -9.9 degrees last season. For now, let’s assume this pitch had the same VAA. With help from our friend Pythagoras, we can calculate that a pitch traveling at an angle of -9.9 degrees would be 1.48 inches higher when it crossed the front of the plate than when it crossed the middle of the plate. Here’s how that works.

OK, so measuring at the front of the plate, the pitch comes in 1.48 inches higher. It’s now missing the zone by just 2.50 inches. It’s well within the shadow zone. That certainly makes it sound a little closer, don’t you think? Here’s what that looks like in our original diagram.

You know what? It’s still pretty far away from the strike zone. Stark’s article mentioned that after the game, reporters told Scherzer that his pitch would have been a strike according to a 3-D zone. They were way off base. In order to do so, the pitch would have had to arrive at the plate with an absurd VAA of 25 degrees. That ain’t happening. This pitch is still unequivocally a ball. There’s no system – goo-based, camera-based, vibes-based, none – in which this pitch hits in the strike zone. It was so far away that Carole King wrote a different song about it, also called “So Far Away.”

That said, I do suspect that this particular curveball actually had a steeper VAA than -9.9 degrees, making it a bit closer than the graphic above indicates. Just using the old-fashioned goo-orb test, it looked sharper than the typical Scherzer curveball. Second, I was talking things over with Michael Rosen, our resident pitching genius, and he got curious and pulled data for a Scherzer curveball, just one random curve from 2023. That pitch had a VAA of -10.1 degrees over the last 10 feet.

That VAA would move Scherzer’s pitch a few hundredths of an inch closer to the zone, and this one solitary, particularly sharp curve could’ve been even closer. It’s still not a strike according to any definition of the strike zone, but it highlights the disconnect between the two current competing versions.

So far we’ve only been talking about the front of the plate, but this would also be true of both the back and the sides. A pitch with a steep horizontal approach angle can clip the corner of the plate before it reaches the midpoint. The back gets tricky because of the plate’s pentagonal shape, but it’s still possible; the closer to the center a high pitch is located, the better a chance it will have of dropping down and catching a piece of the rulebook zone. ABS would tell you that every one of the pitches illustrated below is a ball. But according to a normal three-dimensional strike zone – which is what umpires are calling – that’s not actually true. It’s smaller than the rulebook zone.

As things stand, when the league does implement an ABS challenge system for regular season play – and at this point, that seems like a virtual certainty, though which regular season is still undecided – then the game will officially have two different strike zones. It’s possible that the league could change the rulebook definition for umpires so that it matches the Statcast zone, but that strikes me as unlikely for many reasons, chief among them it would essentially turn the iconic shape of home plate into a vestigial appendage. In the two-zone world – the world that Triple-A players have been living in for a while now – a pitcher would be able to throw a strike, get robbed by the umpire, challenge that incorrect call, and lose the challenge because according to the robot umpire, the pitch really was a ball. Even crazier, the pitcher will throw a strike, the umpire will get the call right, and then the batter will challenge it and that correct call will get overturned! The umpire and the computer will make two different calls, and both will be correct because they’ll have two different zones.

As the numbers from our curveball example show, we’re not talking about a couple of unlikely edge cases. The differences in movement from the front and back of the plate to the middle aren’t minuscule. Some pitchers’ curves average above 11 degrees of VAA, and the sweepiest sweepers average more than six degrees of horizontal approach angle. We’re often going to be talking about well over an inch of difference. This is going to happen all the time. I’m not the first person to notice this. On Wednesday, Baseball Savant’s Tom Tango crunched the numbers and announced that in 2024, one percent of all takes would have fallen into this category, just for issues with the front and back of the plate.

As things look right now, baseball will soon officially have a human strike zone and a robot strike zone. The robot strike zone will be so thin as to be non-existent, while the human strike zone, as it always has, will be shaped like an infinite number of infinitely thin home plates. Honestly, I don’t know how any pitcher who’s had it fully explained to them will avoid succumbing to paralysis halfway through their windup and toppling off the mound simply because they’ve exhausted their ability to process the disjuncture of the situation.

I mentioned earlier that setting up the robo-zone in two dimensions rather than three was a perfectly reasonable choice. The more I think about it, however, the more I think it might be the only reasonable choice. Calling balls and strikes is incredibly difficult. I’ve had to do it before, and I’d approximate that I felt 100% certain on about 30% of the pitches I called. But even then, I doubt I was really thinking about the strike zone the way the rulebook demands. The rulebook zone doesn’t have four corners; it has 10 corners. And it doesn’t have an edge; it has 15 edges. The difference between a two-dimensional plane and a three-dimensional space is the difference between a topographical map and a mountain.

On the one hand, this makes me wish the robo-zone were three-dimensional, just because I’m imagining how much more fun the challenge graphics would be. We’d see in precise cartoon glory not just whether the ball nicked the corner of a box, but one particular corner of a 10-cornered pentagonal prism. It would rule. On the other hand, it’s absolutely preposterous that we ask human beings to process information with anything approaching this level of precision. Wherever you’re sitting right now, try to imagine a pentagonal prism floating in the air next to you. Now try to picture yourself deciding whether a Tarik Skubal fastball nicked one of its seven sides. Now do it again, but first squish your prism down a bit because Nick Madrigal is up next. So maybe it does make sense to have two zones; we’ve just got them reversed.

Scherzer was candid and engaging with reporters, and after processing all of this information, he closed with the takeaway that most of us saw in the headlines: “Can we just play baseball?” he asked. “We’re humans. Can we just be judged by humans? Do we really need to disrupt the game? I think humans are defined by humans.” When he puts it that way, it’s a pretty reasonable request. Right now, umpires and batters track pitches using the exact same equipment, and that makes plenty of sense. If the game is played by humans, it’s certainly not laughable to feel that human eyes and brains should be deciding what’s a strike and what’s a ball. I don’t mean to say that there’s wisdom in every mistake simply because it’s made by a human, but once a computer is making the decisions, the objective of the game becomes slightly less fun, for the same reason that playing chess against the computer isn’t particularly enjoyable. It becomes less of a game and more of a problem solving exercise.

These days, there’s no end to the ways that computer programs are judging us – CAPTCHA requests, Spotify recommendations, suspicious login emails, targeted advertising, personalized search results, automated insurance denials, the artificially indiscriminate firings going on throughout the federal government – and with vanishingly few exceptions, the people being judged would like nothing better than to smash all of these robot judges with a hammer.

Don’t get me wrong, I would love to smash the computers that turned Google into such a joke with a hammer, but the difference here is that many of those systems were designed as shortcuts, either to save time, to replace human workers, or to shift accountability away from the person instituting a crappy policy and onto the circuit board that implements it. On the other hand, the challenge system is a particularly elegant solution to the problem at hand. It will introduce an extra layer of accountability into umpiring without replacing the umpires or undermining their centrality to the game. It won’t obliterate the value of pitch framing, but it will hopefully reduce the amount of shouting umpires have to bear. Now that we have the ability to know the exact location of every pitch, it’s probably not completely defensible to just ignore that knowledge. Instant replay was instituted for the same reason. “I like it when somebody screws up and somebody gets screwed over” is not exactly a winning campaign pitch.

Let me hit you with one last disconnect. The really funny thing is that depending on how you look at it, Scherzer is both the best and worst messenger for this argument. He’s a sure-fire Hall of Famer and a longtime union rep. He’s not afraid of a fight, and his standing in the game ensures that when he speaks, people will listen. His comments warrant plenty of counterarguments, but “Max Scherzer doesn’t know what he’s talking about” is not among them.

On the other hand, Scherzer has never had that much use for umpires in the first place. Since Sports Info Solution started tracking pitches in 2002, 328 pitchers have thrown at least 800 innings. Scherzer’s 14% swinging strike rate ranks ninth among them and his 27% whiff rate ranks 19th. His 17% called strike rate, however, ranks all the way down at 212th. Scherzer has always succeeded by racking up whiffs, pumping his fastball by hitters and tempting them into chasing sliders and curves. Relying on the umpire for called strikes has never remotely been his game. In fact, since 2008, Statcast says he’s had 1,262 would-be strikes stolen from him, third-most in all of baseball. Few players have relied less on human umpires or accumulated more reasons to be fed up with them than Scherzer. Maybe we should tell him that after his next start. I’m sure he’ll have something interesting to say about it.

Last week, I wrote about the careers of the two former college baseball players who have been featured on this season of Love Is Blind, and don’t worry, I’m not going to follow up with a detailed breakdown of their performance on the episodes released this past weekend. (Though if anyone wants the short version: It’s been pretty dire. Ben is getting flamed on TikTok so bad his fiancée is thinking about pulling the plug, while Dave… I don’t know what you’re doing, man. Get it together. You’re in your mid-30s. You should be able to have a frank, productive conversation with your partner.)

I bring all this up because it’s been hard to shake something I mentioned in Friday’s article: Ben Mezzenga’s astonishingly high incidence of taking strike three. In his best years, only about half of his strikeouts came swinging. A typical big league hitter strikes out three times swinging for every time he strikes out looking. Last year, José Ramírez ran a ratio north of 15-to-1, the highest mark in baseball. Cavan Biggio was the only hitter who had 50 or more strikeouts with more than half of them coming with the bat on his shoulder. Read the rest of this entry »

Nobody wants to throw a backup slider. They are, definitionally, an accident. But announcers and analysts alike have noted that these unintentional inside sliders — perhaps due to their surprise factor — tend not to get hit. In 2021, Owen McGrattan found that backup sliders, defined as sliders thrown inside and toward the middle of the strike zone, perform surprisingly well.

I’ll add one additional reason these pitches are effective: They move more than any other slider.

I analyzed over 33,000 sweepers thrown by right-handed pitchers in the 2024 season. I found a clear linear relationship between the horizontal release angle of a sweeper and the horizontal acceleration, better understood as the break of the pitch. On average, as the horizontal release angle points further toward the pitcher’s arm side, the pitch is thrown with more horizontal movement.

Josh Hejka, a pitcher in the Philadelphia Phillies minor league system, told me these results corresponded with his anecdotal experience.

“I’ve often noticed — whether in game or in the bullpen — that the sliders I throw arm side tend to actually have the best shape,” Hejka said. “I believe it’s conventional wisdom across baseball that the backup sliders tend to actually be the nastiest.”

Check out all the movement Corbin Burnes gets on this backup slider from last season.

The relationship between horizontal release angle and movement also holds true for sinkers. When a sinker is aimed further to the glove side — for pitchers facing same-handed hitters, this would be a backdoor sinker — the pitch gets, on average, more horizontal movement, as is the case with this pitch from Anthony Bender.

The explanation for the relationship is straightforward enough. When sweepers are thrown to the arm side and sinkers are thrown to the glove side, the pitcher’s grip is such that maximum force is applied to the side of the baseball, allowing for more sidespin. In a 2015 interview with David Laurila, then-Royals pitching coach Dave Eiland described why sliders back up.

“They really get around it; they don’t get over the top and pull down,” Eiland said. “It’s unintentional, more of a misfire, so to speak. If you could do that intentionally, you’d have a decent pitch.”

It isn’t just sweepers and sinkers that show a relationship between release angles and movement. Back in August, I investigated the mystery of the invisible fastball. Why was a pitch like Shota Imanaga’s fastball, with its elite vertical movement and flat approach angle, so rare? I found that vertical release angles mediate the relationship between both variables. A fastball thrown with a flatter release angle gets less backspin, and so to achieve both requires outlier mechanical skills.

Release angles don’t just measure the nature of a grip, they also dictate the location of the pitch. I conclude that where the pitcher aims a pitch changes the way it moves. For fastballs, pulling down on the ball allows for more backspin. For sweepers and sinkers, getting around the ball allows for more sidespin. Analysts attempt to separate “stuff” from “location;” these findings complicate that conversation.

***

Before we go any further, it’s important to know what exactly is a release angle. Release angles measure — or, in this case, approximate — the angle at which the ball comes out of the pitcher’s hand. For vertical release angles, anything above zero degrees suggests the ball is pointing upward at release; most vertical release angles, particularly for four-seam fastballs, are negative, meaning that the pitcher is aiming the ball downward at release.

Horizontal angles work the same way, but in the x-dimension. Positive values mean the ball is pointed toward the pitcher’s left; negative values point toward the pitcher’s right. (This is a feature of the original Pitch F/X coordinate system, when it was determined that x-dimension pointed to the catcher’s right.) In any case, release angles, both horizontal and vertical, attempt to capture the exact position of the ball at release. Because they capture the position of the ball at release, they contain information about the pitcher’s aim and, it turns out, the force they’re applying to the ball.

My research finds that there is a relationship between horizontal release angles and horizontal acceleration. In simpler terms, the way the ball is released out of the hand, and therefore where it is aimed, impacts the movement of the pitch.

There are some confounding variables in this specific relationship. The Hawkeye cameras (and, in earlier times, the Pitch F/X technology) report accelerations in three dimensions. These accelerations are measured relative to a fixed point on the field, which happens to be right in front of home plate. Because these accelerations are fixed to one point, the reported values can be biased by the position of release in space. This is far from intuitive, so it might be helpful to consider an example.

Remember that Burnes sweeper from the introduction? It accelerated at roughly 16 feet per second squared in the x-dimension. Imagine that instead of throwing his sweeper from the mound, Burnes threw it from the third base dugout. It’s the exact same pitch as before — same velocity, same horizontal break — but the release point has completely changed. On a fixed global coordinate system of movement measurement, the acceleration in the x-dimension no longer describes the pitch’s relevant movement; all that sideways movement would instead be measured in the y-dimension.

This is an extreme example to illustrate the point, but on a smaller scale, this fixed point measurement system biases acceleration measurements. In order to fix this bias, accelerations can be recalculated to be relative to the pitch’s original trajectory, removing the influence of the release point on the acceleration value. These calculations come courtesy of Alan Nathan; Josh Hejka rewrote them as Python code, making my job easy.

A slight nuance:

The accelerations given by MLB (ax, ay, az) are biased by pitch location.

To make these values location-agnostic, we need to adjust the acceleration vector to be relative to the initial trajectory (i.e. the initial velocity vector vx0, vy0, vz0).

?: https://t.co/CVVNKtG3YM

— Josh Hejka (@hedgertronic) November 14, 2024

Even after accounting for these confounding variables, the relationship between release angles and movement is still present. As the plot shows, it isn’t a particularly strong relationship — when modeled, a two-degree change in horizontal release angle is associated with roughly a foot per second increase in transverse acceleration. But while the relationship is not as strong as that between four-seam fastballs and vertical release angle, it is nonetheless meaningful.

Alternatively, the relationship can be measured using good old-fashioned “pfx_x,” or horizontal movement, which is also measured relative to the pitch’s original trajectory. Why go through all this effort to transform the accelerations? For one thing, I had a good time. And also, isn’t it fun to imagine Burnes throwing sweepers from the dugout?

The plot of horizontal location and horizontal movement, with each pitch colored by its horizontal release angle, illuminates the ostensible lack of a relationship between pitch location — measured by “plate_x” on the plot below — and movement. Draw your attention to the patch of dark blue dots around the -2 line of the x-axis. There are two potential ways for a sweeper to end up two feet off the plate inside. It can be thrown with a horizontal release angle around zero and little sideways movement, or it can be thrown with a negative horizontal release angle and lots of sideways movement.

The same relationship holds true for horizontal release angles and two-seam fastballs after the aforementioned adjustments.

On the individual pitcher level, the relationship is slightly weaker; on average, the r-squared is roughly 0.04 for sweepers, with variation between pitchers on the strength of this relationship. Zack Wheeler’s sweeper movement, for example, appears to be particularly sensitive to release angles:

***

Ultimately, analysts attempt to separate “stuff,” defined as the inherent quality of a pitch, from “location,” defined as where the pitch ends up. But what this research suggests is that, to some degree, these two qualities are inseparable. (I wrote about this a bit on my Pitch Plots Substack last September.) Certain pitches generate their movement profiles because of where they’re aimed out of the hand.

These findings naturally lead to deeper questions about the interaction between biomechanics and pitch movement. While there are variables (arm angle, release height, etc.) that are commonly understood to influence movement, these findings suggest that there are even more granular factors to explore.

Is the angle of the elbow flexion at maximum external rotation the most influential variable? Is it hip-shoulder separation? Torso anterior tilt? Pelvis rotation at foot plant? How much do each of these components contribute to pitch shapes?

Thanks to data from Driveline’s OpenBiomechanics Project, it’s easy to model the relationship between dozens of biomechanical variables and the velocity of the pitch. There are about 400 pitches in the database; by attaching markers to a pitcher moving through space, points of interest can be calculated and then compared to the pitch’s velocity.

In this public dataset, Driveline does not provide the movement characteristics of the pitch. But if the force applied to the ball based on the direction of its aim affects the movement of the pitch, it follows that these variables could be measured in a detailed manner. On the team side, KinaTrax outputs provide the markerless version of these data, providing a sample of hundreds of thousands of pitches from a major league population. Imagine the possibilities.

Correction: A previous version of this article misstated the units of acceleration of Burnes’ backup sweeper. It is feet per second squared, not feet per second.

In the middle of the 2024 season, MLB released bat tracking data for the current year. It was a huge revolution in publicly available data, taking something previously observable but not measurable and turning it into numbers. You can see how hard Giancarlo Stanton swings, but now you can also quantify how different that is from other large hitters. Luis Arraez’s superhuman coordination is obvious from watching him play. But in terms of getting his barrel on the ball, relative to the rest of the league, how superhuman is he? Now we know. I think that public research on this front is likely to deliver more and more insights in the coming years.

Of course, what we all wanted to know about bat speed wasn’t available right away. Namely: How does it change? Was Ronald Acuña Jr.’s disappointing start to the season related to an inability to impact the ball with force? Did Matt Olson’s decline have more to do with bat speed or plate discipline? Also, plenty of non-Braves questions, presumably. In any case, we couldn’t say much about that because all we had were the 2024 numbers.

Guess what: Now we have some 2023 data. MLB and Statcast released 2023 data starting after the All-Star break, the earliest data we’ll ever get because that’s when the bat tracking infrastructure got going. Obviously, we’re also going to get more year-over-year data when the 2025 season starts. But our first crack at multiple seasons of data is still noteworthy, so I set out to look through the numbers and came to a few conclusions. I don’t intend for these to be comprehensive, and I’m sure that a measured and careful approach is going to tease out some new insights that I don’t have. But the data came out yesterday, and here are a few highlights.
Read the rest of this entry »

Earlier this week, I published my findings about the relationship between when free agents sign and the size of their contracts. As a quick refresher, in recent years, the last 20% or so of free agents to sign have been settling for contracts meaningfully lower than pre-offseason expectations. But that finding raises more questions, some of which I hope to answer today.

First, there’s an obvious question: Did the free agents who got those late, discounted deals perform worse than expected during the following season? In other words, did their low-dollar-value deals foreshadow lower-than-projected production? To examine this, I took the upcoming season’s projections for the players ranked on my Top 50 Free Agents list in each of the past three years, 150 players in all, to come up with a projected WAR for each segment of players. I then compared it to how they actually did in the ensuing year. There is indeed a drop-off for those who signed late:

First things first: Every group underperformed its projections. That comes down to playing time. Our projections use Depth Charts playing time, which approximates the most likely distribution of playing time across a given roster without accounting for the likelihood of injuries. Just as an example, non-catcher batters were projected for an average of more than 600 plate appearances in this dataset, and they came in closer to the mid-500s in practice. So don’t pay too much attention to the absolute numbers; the relative differences are what to look at here.

The last 10 free agents to sign saw huge shortfalls in production relative to expectations. One reason: They played less. The average hitter in this group of 150 free agents batted 70 times less than projected. Hitters signed among the last 10 free agents in their class batted 100 times less than projected. Likewise, the average pitcher in the group came up 25 innings shy of projections, but pitchers among the last 10 players signed came up 40 innings short.
Read the rest of this entry »