This is a collaborative effort made by Ethan Rendon, Elijah Emery, Will Sugar, and Tieran Alexander.
There is a common misconception that pure stuff and command are all that determine performance. That analysis is obviously missing something or Matt Brash would be one of the best starting pitchers in baseball. He throws an 80-grade slider in terms of specs with three more great pitches. His problem was not throwing strikes; Brash actually found the strike zone at an above-average rate out of the rotation. Yet, Brash posted a 7.65 ERA and 20 K% in that role. Stuff and command can’t be the only thing that defines pitcher performance.
There is a missing ingredient to his struggles: pitch tunneling. Theories about tunneling pre-date the second dead-ball era. Despite the ideas, we’ve considered it something not quantifiable. Most people see pitch tunneling as a lazy hand-waving explanation for stuff fans don’t understand. As the rest of pitching analysis has become quantifiable, pitch tunneling has remained one of the few concepts left in the dark.
The word impossible is only found in the dictionary of fools. No one has publicly quantified pitch tunneling yet, in a way that actually explains performance. That is the only conclusion we can draw- not that tunneling is unquantifiable. Pitch tunneling is the difference in how pitches break before they reach the plate; pitch movement data from statcast has been publicly accessible since 2017. So why has tunneling been so difficult to quantify?
Previous Works on Tunneling
Baseball Prospectus published the first pitch tunneling metric for public consumption in 2017. They attempted to quantify the separation between two pitches based on the separation between them when they were 24 feet from the plate. They also factored in the difference in release between two pitches and considered pitch sequencing. The model might have been needlessly overcomplicating things because back-to-back pitches aren’t what determine swing decisions in the modern game, it’s what the pre-game prep work says.
If you’ve ever seen two pitches cross paths, as they approach the plate and perceive it as pitch tunneling, you’re not alone. Videos and GIFs of pitch overlays often go viral, implying that batters can be fooled easily by the aesthetically pleasing movement. However, this viewpoint is largely inaccurate. The hitter’s perception defines pitch tunnels, not what we see on TV. The other problem is the sample size. Two pitches might look similar at one point during their flight, but that does not mean they are similar the entire time. Pitching works on the law of average. Batters don’t react to a specific two-pitch pairing, but based on what they think the incoming pitch will do; more often than not determined by their pre-game preparation in the modern age. Instead of looking at just a two-pitch sequence, perhaps it is best to view pitch tunneling the same way hitters game plan; looking at the entire arsenal and pitch averages.
Why Release Point Separation DOES NOT MATTER
On a similar note, fans like to look at release point consistency when determining if a pitch tunnels. If a player is throwing the slider from a release point that is four inches lower than the fastball, one could expect that the pitches betray the pitch type before he even throws the pitch. In theory, that makes sense, but release points are ranges. Shohei Ohtani is that pitcher I just described. His release averages are inconsistent, but based on the chart below, it is obvious that the two pitches overlap, even if the averages don’t. Consistently inconsistent release points are just as effective for deception as consistent ones.
Our Version of Tunneling - Simplifying The Concept
Release points, bad GIFs, and pitch sequencing are the usual methods of evaluating pitch tunnels, but we took a different approach. Instead of looking at a 50 millisecond window, we zoomed out and evaluated the entire pitch path. We operate based on the law of average, rather than one pitch at a time, and have identified the ideal ranges of each element of pitch data and the ideal differences in shape from the primary pitch to elicit out-of-zone swings.
The goal of tunneling is to generate swings at unhittable pitches. Pitchers endeavor to trick the batter into swinging at a slider in the dirt by making them think it’s a fastball. Thus, we believe chase rate is the best way to measure how often the pitcher deceives the batter. Using raw swing rate rewards pitchers who throw obvious meatballs that encourage in-zone swings, while using whiff rates reward hitters who aren’t swinging. Chase rates aren’t the perfect method– they still rely heavily on stuff– but are a solid proxy for estimating pitch tunneling ability.
This is a baseline to prove that pitch tunneling can be quantified. It is not the only way to evaluate the ability to tunnel pitches. There are exceptions to every rule, and this is not to be taken as the gospel. Models have a limited perspective that forces everyone to fit in a cookie cutter mold and cannot understand the give and take relationship of pitch movement and velocity.
Sliders
The four-seam fastball and the slider are the most obvious pitch pairing for noticing the impact of pitch tunneling. There are tons of disgusting sliders that don’t get out-of-zone swings. Similarly, there are tons of gyro sliders that don’t look like all that much, but consistently dominate because they play off of the fastball so effectively. We found three basic rules a fastball and slider pairing usually follow when they regularly get out-of-zone swings.
6-14” Horizontal Separation
8-16” IVB Separation
6-11 MPH Separation
Based on our studies, we’ve found that almost every effective slider falls into at least two of those three boxes. Sliders can have too much horizontal movement as the graph above shows. Separation in the 6-14” range is most ideal for getting swings out of the strike zone. Anything over 18” of separation is borderline unusable. The horizontal separation split can work with more movement on the fastball, or more movement on the slider.
Justin Verlander has an excellent horizontal tunnel on his fastball/slider pairing with 10” of run on the fastball and 4” of sweep on the slider. Drew Rasmussen has an effective tunnel that gets chases with only 2” of run on the fastball and a sweeper with 12” of horizontal to pair with it. Either combination works; it’s the separation that matters, not just the individual pitch’s shape.
On the vertical axis, a pitch may also have too much or too little movement. Once again, we want to focus on the separation from the fastball and slider rather than raw pitch movement figures. Ideally, the slider has 8-16” IVB of separation from the four-seam. This can once again go in either direction with more IVB paired with very little lift, or less IVB paired with more depth. Anything approaching over 20” of separation can be completely ineffective, though some of the results in our graphs are a byproduct of selection bias. Most sliders with that much depth are sweepers or slurves with a ton of horizontal movement.
Velocity also plays a key role in defining pitch tunnels. Our research found that ideally, a pitcher throws his slider no more than 10.5 MPH slower than the fastball. This makes sense on paper. A slower pitch will often break earlier because of the lack of speed. Hitters are not the best at gauging the speed of a pitch, but they are competent enough to recognize vast gulfs in velocity. We’ve also found that anything less than 5.5 MPH slower than the fastball is pushing it. This is again logical because the goal of throwing secondary stuff is often to disrupt the hitter's timing. The problem with a slider as hard as the fastball is that all the batter has to do is correctly guess the location to hit it.
So what happens when a pitcher has a nasty slider but no tunneling ability? Look no further than Jeff Hoffman for the answer to that question. Hoffman has a slider that looks unhittable on paper. He throws it at 81.2 MPH with -4.8” IVB and 10.9” of horizontal movement. That is a slider that projects to miss a lot of bats and, for the most part, it does with a 38.5% whiff rate. The problem is that he has 21.6” of horizontal movement separation from the fastball and 21.9” of IVB separation. There is also 13 MPH of velocity separation between the two. Because of those factors, an elite slider in theory has an 18.8% chase rate in practice. His tunnels weren’t good last year either, but adding 5” of horizontal separation has dropped the chase rate by 10% despite the stuff grading out higher on a stuff model.
Even mediocre stuff can become exceptional when it checks all three boxes for a perfect tunnel. Look no further than Tony Gonsolin. Gonsolin might win Cy Young if it wasn’t for another tunneling wizard in Sandy Alcántara. Gonsolin has zero pitches that grade out above a 50 on stuff according to the pitching bot. Yet, the slider is one of the most dominant pitches in baseball. The stuff isn’t there, but the tunnel is immaculate. Gonsolin has 12.7” IVB separation, 11.7” HB separation, and only throws the slider 5.7 MPH slower than the fastball. Because of his perfect tunnels, he has a 37.6% chase rate on a pitch that grades out as below-average based on raw stuff. That is how game changing a good slider tunnel can be.
The Miracle Pitch
Most of the players with an extreme deviation in their horizontal movement on the fastball and slider who still find results have one thing in common: they throw the cutter. In fact, half of the sliders in our dataset that have over 17” of horizontal separation from the primary and still get chases 35% of the time pair with a cutter. Cutters are the living miracle that make sweepers usable and tailing two-seamers effective as well. The purpose of a cutter is to split the plate and bridge extreme gaps in both vertical and horizontal movement between a fastball and slider.
Yu Darvish is perhaps the best example of a bad tunnel being made good by the existence of the cutter. Yu Darvish has 22.3” of horizontal separation between the fastball and slider, which should not work at all. The cutter is by far Yu Darvish’s worst pitch in terms of results with a .372 wOBA that has most Padres fans screaming that Darvish should never throw it. His slider and fastball get much better results. However, it is only because of the cutter that eliminates his bad horizontal tunnel and gives him optimal horizontal separation from the fastball and improves his vertical tunnels. It also bridges the velocity gap that is too extreme between the fastball and slider. The cutter isn’t good in a vacuum, but he needs it to make the rest of his arsenal work.
Evan Phillips is a testament to how much better the Dodgers are at tunneling than the rest of baseball. He was terrible in Baltimore and continued to struggle when he first joined the Dodgers. This year, Phillips has been one of the best relievers in MLB. The slider has 21” of horizontal movement separation from his fastball, and got no chases prior to 2022. He added a cutter, putting his slider and fastball within the ideal ranges to perform. The fastball and slider don’t have to look the same, they just have to look like the cutter, which looks like both of them.
The easiest solution to a lack of chases on the slider with filthy stuff is usually to add a cutter. If, for instance, Matt Brash added a gyro cutter with average shape, he would likely have elite chase rates on both the sweeper and fastball as it would bridge the gap between their movement profiles on both planes and help the velocity separation that would most likely re-emerge as a starting pitcher. (Throwing harder has had a similar effect out of the bullpen as it has led to less horizontal movement on all of his pitches.)
Curveballs
Curveballs are incredibly complicated. There are two variants of curveballs, those that prioritize chases and those that prioritize called strikes. The slower curve with more movement gets more called strikes because it fools the batter into thinking it is not a hittable pitch, as they watch the ball float into the zone. As long as a pitcher is throwing a slow curveball in the zone, it will get called strikes.
The other variant of curveballs is the chase ones that share a tunnel. Those pitches are often hard curveballs with minimal vertical and horizontal separation from the fastball. They get a lot more in-zone swings, but also a lot fewer called strikes. Curveballs are a give and take relationship.
We believe that curveballs that draw chases are the better variant of curveball. As illustrated above, curveballs that get chases produce lower wOBAcons than those that don’t. Curveballs that get chases also tend to get higher called strike rates on their fastball. This is an interaction unique to the chase curveball. The chase rate on a slider has little to no effect over fastball performance, and while changeups have an effect it’s not for the same reason. The curveball might not be any better in a vacuum if you prioritize tunneling over making it standalone and get called strikes, but it sets up the arsenal better because it enhances the fastball and gets weaker contact, resulting in easy outs. The benefit of the tunnels isn’t the same as the slider, but there is a tangible benefit that makes a clear impact in games.
Changeups and the Value of Command
Changeups don’t behave as any other pitch does, largely in part because they move to the same side of the plate as fastballs so can’t naturally build tunnels in the same manner. Chase rate can be predicted, albeit it has less to do with how it separated from the fastball and more to do with pure stuff and location. There’s also some stuff that is much less easily quantifiable and not remotely graphable, such as arm speed and feel that clearly plays a role in getting a changeup to perform.
The single most important part of a changeup's ability to draw chases is the VAA separation from the four-seam fastball. This is partially due to needing to maximize drop on the pitch off of the rising fastball, but it goes deeper. Location is actually the key factor and why we use VAA instead of vertical separation. Using the fastball up and the changeup down will in most cases let the changeup perform above expectations and draw more chases than it might otherwise get.
Changeups also don’t really care about horizontal movement separation from the fastball. This is because all changeups naturally share a tunnel with the armside heater due to them moving in the same direction. What does matter on a changeup, however, is raw horizontal movement. Typically, anything over 15” is most effective.
Velocity separation also plays a role, but that is more for in-zone whiffs than getting chases and isn’t strictly a byproduct of tunneling. Velocity separation is also only beneficial when a pitcher is capable of regularly throwing his changeup in the strike zone. Since all changeups tunnel, a pitcher should aim to maximize changeup stuff and optimize their locations. The fastball shape is less important to how a changeup might play than the fastball location. Changeups might not have a significant correlation in terms of tunneling, but that’s largely due to the fact that every fastball and changeup pairing tunnels; there are no bad changeup tunnels.
Sinkers
Sinkers have a lot in common with four-seamers, but there is enough difference in their behavior that the primary pitch type affects what the ideal tunnel is. The sinker has a slightly wider range for ideal horizontal movement with the slider because the sinker is a more naturally tailing pitch with closer ranges of effective velocity because of the extreme tailing action. Sinkers function best in a horizontal tunnel with up to 16” of horizontal separation. In addition, they function with a slightly wider range of vertical movement- anything between 6-18” of vertical separation. There appears to be almost no effect from the actual velocity in the sinker/slider tunnel.
Sliders work best with pseudo two-seam fastballs, and bowling ball sinkers that minimize horizontal movement. Curveballs are the opposite. With a curveball, an elite sinker that maximizes horizontal movement while minimizing the vertical separation element sets the better chase curveballs up. Slurves fall somewhere between the slider and curve, with ideally them having minimal horizontal separation and elite depth on the sinker.
Changeups function very similarly off of a four-seam and sinker. The sinker needs slightly less vertical separation to be considered good, but you still want to maximize vertical separation to make the changeup work in a vacuum. A sinker/changeup also likely needs more horizontal break to make up for having a less prominent VAA difference. Essentially, the best version of a sinker for a changeup tunnel is a two-seam fastball with tail and not a ton of drop. Maximizing velocity separation is how you create an effective tunnel off of a good sinker as those ones will rack up in-zone whiffs.
Secondary Interactions
The fastball is the most common source of a pitch tunnel but it’s not the only one that can tunnel. Pitchers like Kevin Gausman have an excellent slider/splitter tunnel that allows both pitches to play up. The goal of a secondary tunnel is typically to minimize one of horizontal and vertical separation while maximizing the other.
Oftentimes this will be like the Gausman example, who has merely 2.3” IVB separating the two pitches but 16.4” of horizontal separation. To make the tunnel even more effective, Gausman is throwing the two pitches at the exact same average height of 1.48’ and to opposite sides of the plate. The best performances of Gausman’s career have all come with higher slider usage for this reason, even if the slider is a much worse pitch in a vacuum. You don’t see as much impact from secondary tunnels because pitchers rarely throw enough of them to make a huge impact, but they do have a minimal effect as well.
Applications to Player Development
Pitch tunnels can dictate player development. The Dodgers correctly recognized that Andrew Heaney was getting too much horizontal separation between his fastball and sweeper. They switched him to a gyro slider this year, and he’s been arguably their most dominant pitcher while he was on the field. The extreme HAA on the slider has made an otherwise generic gyro slider elite and has fixed all of his tunnels so it now gets chases 48% of the time.
Another example of using tunneling to gain an unfair advantage also comes from another current Dodgers ace. When the Mariners traded for Tyler Anderson last year, they had him start using the fastball in the middle of the plate more- a completely counterintuitive strategy. This saw his cutter and to a lesser extent the changeup draw significantly more chases. The Dodgers saw how much better the middle-middle fastball set up his arsenal and just had to make a few tweaks to his pitch usage- a 8% bump in cutters against right-handed hitters and the same bump in changeups against lefties. The four-seam sets them both up but it didn’t have to be the primary to function. From their understanding of pitch tunnels, the Dodgers got another top starter for just $8 million in free agency.
The impact of pitch tunnels goes deeper than just buying free agents cheap, though. Understanding pitch tunnels helps teams understand how to get the most out of pitchers without earth shattering stuff. Understanding pitch tunnels can help determine how to develop the high schooler taken in the 14th round with a plus fastball and nothing else. If they have 20” IVB then the answer is probably to focus on a gyro slider instead of a curveball. If they have a sinker with crazy depth and run then it’s probably better to focus on developing a curveball instead of a slider. The perfect breaking ball shape can be honed even more carefully when considering things like arm slot- that is how the Dodgers realized Heaney would be perfect with the gyro slider. From the beginning, there is some benefit to choosing the pitches to focus on with tunneling in mind.
Balancing Stuff and Pitch Tunneling
Pitch tunneling will rarely do everything. There is still value in having great stuff that defines a large part of pitcher performance. There are the occasional arms who thrive with “average stuff”- (almost all of whom play for the Dodgers), but most players will need good stuff to be successful. A team should never intentionally neglect stuff to perfect a tunnel. The solution for Matt Brash isn’t to scrap the sweeper, rather to add a cutter, improving stuff and tunnels at the same time. Teams and players have to balance deception and stuff to find success. We know pure stuff doesn’t work without any tunnels, but we also know deception can’t win by itself or someone like Zach Plesac would be a lot better than he is. Pitchers have to find a way to balance both parts of pitching or else they will never reach their full potential.
There is still a lot of work to be done in understanding pitch tunneling. This is not the end of the journey, but one of the earliest stops in understanding a phenomenon that has ruled the game we love for decades. Hopefully, this newfound understanding of pitch tunnels will help someone build a model to effectively predict chase rates for each pitch. If we can build that model, then projecting which minor league arms will translate might be a hundred times easier. Pitch tunneling has changed baseball, and if the Dodgers consider it the key to building an all-time great pitching staff, perhaps the rest of the league should start considering them as well.