Avoid the Dead Zone: An Extensive Analysis of the Relationship between Fastball Stuff Characteristics and Utility through Four Logistic Regression Models

Introduction

The term “stuff” has frequented the art of pitching for decades. In the ball-flight tracking era, stuff has transformed from an optical observation to a quantifiable science. We’ve all seen the electric Jacob deGrom fastball and the air-bending Devin Williams changeup; perhaps we don’t need advanced programming languages to tell us that those are effective pitches. Nonetheless, the days of evaluating pitches by their visual aesthetic have come to a close. It is time to shift our attention to mathematical models that transform bare pitch data into meaningful insights about the utility and strength of a given pitch. 

The success of a pitch is dependent on two variables: the stuff profile and the pitcher’s command of the pitch. To the dismay of many old-school baseball fans who preach the importance of dotting the corners, pitchers with elite stuff are garnering the attention of professional teams far more frequently than the artists of command. That isn’t to say command holds no value, it is simply less quantifiable and, more importantly, less correlated with pitcher success. Not only is stuff sexy, but stuff also wins baseball games. 

It is probably important to define what I mean when I say stuff. In short, the term “stuff” represents the three-dimensional shape of the ball’s trajectory from the pitcher’s hand to home plate, as well as the time it takes to travel this distance. This shape is influenced by quantifiable variables, such as velocity, spin, and release point. Velocity is not only a determinant of the batter’s reaction time to hit the ball, but also a determinant of the effect gravity has on the baseball. The faster the pitch, the lower gravity’s ability to weigh down the baseball. The magnitude and direction at which the ball is spun, along with the seam orientation, determine the induced vertical and horizontal movement on the baseball. Last but not least, the three-dimensional release point heavily influences the shape of the pitch’s trajectory, considering all pitchers are aiming at the same strike zone regardless of where the ball starts. 

There is a ton of value in boiling down pitch characteristics into a singular metric that quantifies how strong a pitch is, but a Stuff+ model fails to detail the intricacies of a pitch’s utility. Trust me, there will be articles to follow about my designing of a BaseballCloud Stuff+ model, but this article is headed in a slightly different direction. Fastball A and Fastball B could have identical stuff grades, but could serve entirely different purposes in a pitcher’s arsenal. One could be a rising fastball that is extremely effective in two-strike counts, whereas the other could be a sinker with a ton of seam-shifted wake that generates a ton of ground balls. Are these pitches equal in value? Possibly. Are they equal in utility within a pitcher’s arsenal? Most definitely not.  

To put it bluntly, this article is an analysis of the relationship between fastball stuff characteristics and fastball utility. Instead of arguing that Fastball A is good or bad, the goal of this article is to tell you which domains of pitching Fastball A should dominate and which domains Fastball A should struggle. After all, once the pitch design sessions are over and the stadium lights turn on, it is far more valuable for a pitcher to know the potential utility of their fastball as opposed to a bare stuff grade. 

Fastball Utility

I believe that pitches serve four purposes. The first—and the one that I believe is most reflective of the strength of a fastball’s stuff—is the ability to generate a whiff in the strike zone. In a 2-0 count with the bases loaded, the pitcher knows they need to throw a pitch in the zone, and for most pitchers, the pitch that they have the most command of is their fastball. Unfortunately, hitters know this too. When both the pitcher and hitter know what is coming, who is going to win the battle? That is why it is so important to be able to generate in-zone whiffs on a fastball.  

The second attribute is the ability to generate swings outside of the strike zone. You may be wondering why I chose to say swings instead of whiffs here, and the reason is because any contact that is made outside of the strike zone is likely going to be far weaker than contact made on a pitch in the zone. Take the aforementioned situation, but make it an 0-2 count instead of a 2-0 count. What’s the goal here? To induce a swing on a pitch outside of the strike zone. If a pitch has an exceptionally strong ability to generate chases, it could even work in the 2-0 count. 

The third purpose of a pitch is to generate called strikes. In fairness, this is less correlated with stuff and more correlated with command and hitter approach, but I would be lying if I said stuff had nothing to do with it. At the end of the day, high velocity pitches provide the batter with less time to react than a low velocity pitch. This is also an appropriate time to introduce the idea of approach angles. Vertical and horizontal approach angles have gained a ton of buzz in analytics circles as of late. In short, they are simply the vertical and horizontal angle at which the pitch approaches home plate. Low release, high velocity fastballs approach the plate at a flatter vertical angle, whereas high slot, low velocity fastballs take a steeper trajectory. These angles will likely play a strong role in a fastballs’ ability to steal a called strike. 

Although the first three purposes revolve around strike generation, the last one is a bit more old-school: the ability to generate weak contact. The stubborn data analyst in me will forever argue that the best way to limit hard contact is to limit contact period, but the truth is some pitch characteristics seem to influence a hitter’s ability to do damage on a batted ball. While the sinker-ballers of the world may not shine as brightly in the strike generation categories, a strong sinker should induce a plethora of ground balls. 

Methodology

Now it is time to briefly dive into the methodology of my analysis. Using a dataset of over 100,000 NCAA Division I four-seamers and sinkers thrown on Yakkertech systems, I trained four different logistic regression models that quantify a fastball’s ability to serve each of the four purposes stated above. I incorporated the following stuff-based metrics into my models—if you are unfamiliar with any of these stuff metrics, there is a glossary at the end of the article. 

  1. Velocity 
  1. Induced Vertical Break 
  1. Horizontal Break 
  1. Vertical Approach Angle 
  1. Horizontal Approach Angle 
  1. Release Extension 

One instant shortcoming of the models that I foresaw was the assumption that more movement is automatically better. Movement, along with other variables, is rather complex, in the sense that it isn’t always beneficial to increase quantity. Although it is often valuable for a pitch to have a ton of induced movement, there is something valuable about throwing unique pitches. A fastball with slightly below average movement may not benefit from leaping into the average tier. Avoiding the type of fastball that is often referred to as “generic” or “dead zone” is extremely important. Fastballs that fall within the red boxes in the plots below often receive the label of generic. Unless they possess some sort of abnormal quality such as outlier velocity or late break, these fastballs often perform poorly. 

Fastballs, Release Point, Baseball Stuff

To solve this potential shortcoming, I found the NCAA median of induced vertical break, horizontal break, vertical approach angle, horizontal approach angle, release height, and release side, and calculated the absolute value of a fastball’s deviance from the median of each variable. In other words, I created a variable that quantified the genericness of a fastball’s shape. 

Purpose Metric 
Generating In-Zone Whiffs Middle-Middle Whiff% 
Generating Out-of-Zone Swings Chase% 
Generating Called Strikes Middle-Middle Swing% 
Limiting Hard Contact Barrel% on Middle-Middle Fastballs 

In-Zone Whiff Analysis

Although my holistic goal is to find the stuff qualities that best predict in-zone fastball whiff rate, the strike zone is rather large. Command within the strike zone plays a rather significant role in determining the likelihood of inducing a whiff. After all, a pitcher is much more likely to receive a favorable result by dotting a fastball on the corner instead of leaving it down the middle. To adjust for command, I decided that a more refined measure of a fastball’s ability to generate whiffs in the strike zone is to shrink the strike zone to the pitcher’s most vulnerable spot: Zone 5. Major League Baseball’s Statcast system designates pitches right down the middle as Zone 5 pitches, so I will do the same. The logistic regression model that I created outputs expected middle-middle whiff rate based on stuff characteristics. Let’s dive into some of the results. 

Variables that are most indicative of expected middle-middle whiff rate: 

  1. Velocity 
  1. Vertical Approach Angle Deviance from Median 
  1. Induced Vertical Break 
  1. Horizontal Break 
  1. Release Extension 

As I previously stated, I believe that my expected middle-middle whiff rate model would strongly correlate to a Stuff+ model. Since we have adjusted for command, the ability to generate a Zone 5 whiff on a fastball is almost solely a result of stuff. That statement is backed by the fact that 58 of the top 60 fastballs on the model’s leaderboard belong to a certain Tennessee fireballer who touched 105mph out of a 4.6-foot release height. In fact, the fastball with the highest expected middle-middle whiff rate among those thrown on Yakkertech systems in 2022 is the renowned Ben Joyce 105.5mph fastball. If thrown in Zone 5, a pitch with those stuff characteristics is expected to generate a whiff on 59.7% of swings. The highest non-Joyce Fastball sits at 45%.

 

Yakkertech, fastballs, fastball, MLB

Maximizing fastball rise is instrumental in generating middle-middle whiffs. It is no coincidence that induced vertical break and vertical approach angle both play a significant role in the model, as high induced vertical break prevents the ball from dropping, which helps create a flat VAA. Although horizontal break and horizontal approach angle are not as intertwined with middle-middle whiff rate, they are not insignificant. At the end of the day, generating a whiff in the most vulnerable part of the strike zone is not easy, and adding a horizontal element to a fastball is certainly beneficial, especially if the fastball lacks an impressive vertical profile.  

One important finding from the plots is that sinker-ballers do not fare well in generating Zone 5 whiffs. Good sinkers usually fall below 10 inches of rise with 15 inches of run or more. Such a profile is beneficial in certain elements of pitching that we will explore soon, but missing bats is not one of them. This particular facet of pitching appears to benefit those with more induced break, especially on the vertical side. Even generic fastballs appear to play better than those with less-than-average movement in this regard. 

At the end of the day, velocity reigns supreme. Throwing gas is the best way to dominate the most hitter-friendly area of the zone. Among the fastballs that fall in the top 10% of expected middle-middle whiff rate, only 4.3% are thrown at a below-average velocity. Among the fastballs that fall in the bottom 10% of expected middle-middle whiff rate, only 4.4% are thrown at an above-average velocity. Aside from limiting the hitter’s time to react, throwing the baseball at a high velocity also helps maintain a flat vertical approach angle, as the effect of gravity is diminished. Velocity has always been and will always be the defining metric of fastballs worldwide, and for good reason. 

Out-of-Zone Swing Analysis

Before diving deep into a fastball’s ability to generate swings outside of the strike zone, I should clarify that the vast majority of pitchers rarely intend to use their fastball as a chase pitch. There are very few college baseball pitchers—and not many MLB ones, for what it’s worth—who see a rise in fastball usage while ahead in the count. Conventional sequencing still plays a role in all levels of baseball. Pitchers attempt to use their fastball early in the count to get ahead before putting the hitter away with their breakers or offspeed pitches. With that said, “pitching backwards” has become more prominent in the analytics era, and there are pitchers who frequently take the reverse approach, especially if their fastball has chase tendencies. Limiting predictability augments the effectiveness of an arsenal, so throwing a fastball that can be effective in two-strike counts is rather valuable.  

Variables that are most indicative of expected out-of-zone swing rate: 

  1. Velocity 
  1. Vertical Approach Angle 
  1. Release Side Deviance from Median 
  1. Release Extension 
  1. Release Height Deviance from Median 

Unsurprisingly, the results are relatively similar between the expected middle-middle whiff rate model and the expected out-of-zone swing rate model. Limiting reaction time is particularly important when provoking a hitter to swing at a pitch that they should not swing at. Flat vertical approach angles are especially important for inducing swings on fastballs above the strike zone. Hitters expect the ball to drop into the zone, but the backspin and velocity prevent it from doing so. Two variables that are more significant in this domain compared to the first model are release side deviance from median and release height deviance from median. In other words, how generic is your release point? To make a hitter do something they shouldn’t, some element of the pitch must be unorthodox. In many cases, an unconventional release point can be the starting point of a unique fastball shape. 

Yakkertech, Fastballs, fastball

Before diving into the movement profile plot, I first want to discuss the role of angles in generating chases—specifically, horizontal approach angle. Horizontal approach angle does not receive nearly as much chatter in the baseball analytics world as vertical approach angle, but that does not mean it is insignificant. Those with extremely steep HAAs on their fastball generate a ton of swings outside of the zone. Similar to release point deviance from the median, one of the main draws in throwing a fastball with a steep HAA lies in its uniqueness. Fastballs with steep HAAs often come from abnormally wide release points, which makes hitters question how much the pitch will travel horizontally. Another notable observation is that fastballs with a flat horizontal approach angle are more successful than those with a generic angle. The HAA hierarchy goes as follows: steep, flat, generic. Avoid the dead zone! 

The movement profile plot reveals a key difference between fastball stuff in relation to in-zone whiffs versus chases: sinkers. Although sinker-ballers failed to generate whiffs in the zone, they evidently have a strong ability to succeed outside of the zone. It appears as if the main way for a fastball to generate chases is to maximize either the induced vertical break or the horizontal break on the pitch. Generic fastballs and low-spin-efficiency fastballs fail to do either. As I stated in the introduction, the goal of this piece is to analyze the relationship between fastball stuff characteristics and fastball utility. Simply analyzing chase-inducing ability is a bit too vague. We need to learn where these chases are generated. 

Yakkertech, fastball, fastballs

The extent to which these graphs are polar opposites is astounding. Where there is blue on one there is red on the other, and vice versa. Evidently, sinker-ballers make their money at the knees and below, whereas rising fastballs should live in the upper third and above. This concept makes sense in theory as well, considering rising fastballs stay above the barrel, whereas sinking fastballs drop beneath the bat. Attempting to take the opposite approach will not only limit chases, but also catalyze barreled balls.  

These plots also demonstrate that there is a range of fastball movement profiles that fail to induce a ton of chases on fastballs both above and below the zone. Scroll up if you must, but it is important to recall a plot from earlier in this piece that illustrates the distribution of fastball movement profiles at the NCAA Division I level. Notice the overlap between the generic zone on that plot and the white zones on both of these plots? So did I. Some fastballs induce chases above the zone; others induce chases below the zone. Generic fastballs do neither. 

Called Strike Analysis

The reality is that a pitcher’s ability to generate called strikes is as much a product of their command as it is their stuff. Stuff is a factor, as we will see shortly, but throwing a pitch on the edge of the strike zone is the best way to make a hitter take a pitch they shouldn’t take. With the rare exception of auto-take situations, called strikes occur when the hitter expects a pitch to travel outside of the strike zone, when instead it ends up in the zone. The simplest way to do so is to throw a borderline pitch that catches the corner. The closer a pitch is to the notorious Zone 5, the harder it is to trick the hitter into not swinging.  

That is why I decided the metric that best represents a pitch’s ability to generate called strikes purely based on its stuff is expected middle-middle swing rate. Similar to the in-zone whiff model, if a fastball is strong enough to make a hitter not swing in the most hitter-friendly part of the strike zone, then it is a strong called strike generator. 

Variables that are most indicative of expected middle-middle swing rate: 

  1. Vertical Approach Angle 
  1. Velocity 
  1. Vertical Approach Angle Deviance from Median 
  1. Release Extension 
  1. Induced Vertical Break. 

For starters, the best way to force a hitter into making a poor swing decision is to limit their reaction time as much as possible. How does one do that? Velocity. Minimize the time the hitter has to decide whether to swing or not. Vertical approach angle is also largely influential in middle-middle swing rate, likely for two reasons. First of all, high velocity creates flat VAAs, so the two variables are intertwined. Second of all, flat vertical approach angles come with an illusion that makes the hitter think the ball will drop more than it does, but as we discussed earlier, the ball’s velocity and backspin prevent it from doing so. It is very possible that a hitter expects a fastball with an abnormally-flat VAA to drop below the strike zone, but it instead ends up down the middle. This may sound extreme, but the power of velocity and backspin in preventing drop is stronger than some may think. The fact that the five most influential variables revolve around limiting reaction time and preventing drop provides strong evidence that those are the two most effective methods of inducing a middle-middle take. 

Yakkertech, fastballs, strike zone

As one might expect, movement in sheer quantity is vital in called strike generation. To provide the deception of a pitch ending up in a different location than the hitter expects, it is of the utmost importance to induce movement. We can see that middle-middle called strike generation is not a game built for sinkers. Induced vertical break is more significant than horizontal break in this case, and sinkers play the game of killing rise while gaining run. It is also not a game built for low-spin efficiency fastballs. Generic fastballs have more success than below-average movement fastballs in called strike generation. Although it is not always the case, it does seem as if more movement is always beneficial in called strike generation. The dead zone is not as much of a worry as it is in other facets of pitching. 

As I theorized, extremely flat vertical approach angles thrive in called strike generation. VAAs flatter than -4° are called strike savants, as the hitter fails to recognize how little the ball will drop. It also appears that non-generic horizontal approach angles help deceive the batter as well, which could be due to the same reason as vertical approach angles. The physics-based difference between vertical approach angle and horizontal approach angle is the presence of gravity. Because gravity plays a role in weighing the baseball down, velocity is a core component of vertical approach angle. However, gravity does not operate on the horizontal plane, so horizontal approach angle is only a product of the width of a pitcher’s release point relative to the center of the rubber and horizontal movement—location also plays a role, but I aim to control for that when focusing solely on a pitch’s stuff profile. A steeper HAA is better for called strike generation, as it disrupts the hitter’s ability to gauge how much the ball will travel on the horizontal plane, but very flat HAAs once again appear to perform better than generic HAAs. 

To demonstrate the importance of throwing an unorthodox fastball, let us refer to a Big Ten right-hander. Let’s call him Mark. Mark throws a fastball that ranges from 82-86mph, which is well below Division I average. If you think the velocity is unimpressive, wait until we dive into the movement profile. On average, Mark’s fastball gains 11 inches of rise and 11 inches of run. This is a textbook example of a slow fastball with a generic movement profile that is destined to have little success against other Big Ten teams. Or is it? Mark’s release height is approximately 5 feet, which is lower than most but not unusually low. However, he, as a tall righty, starts on the third-base side of the rubber and releases the ball from a sidearm position. He releases the ball over 5 feet to the right of the center of the rubber. His horizontal approach angle is usually steeper than 4°. As a result, he finds himself near the top of the expected called strike generation leaderboard, despite having a well-below-average fastball in all the other stuff categories.  

Barrel Analysis

Before diving into the types of fastballs that reign supreme in the art of contact management, let me first define what a barreled ball is. My classification is slightly different than that of Major League Baseball. In order for a batted ball to be considered a barrel, the batted ball must be hit at an exit velocity greater than or equal to 95mph, and at a launch angle greater than or equal to 8°, but no higher than 32°. To provide context, approximately 17% of batted balls at the NCAA DI level fall into this barrel category. Another important element of the model to note is that quality of contact is also a product of pitch location. Once again, I have decided to limit the model’s training dataset to only batted fastballs in the aforementioned Zone 5.  

I suspect that contact management on fastballs will favor different elements of stuff than whiff generation. Whiff generation revolves around missing the bat entirely. Contact management is a product of preventing the batter from hitting the ball hard while forcing them to hit it at a pitcher-friendly angle. Since Barrel% requires a batted ball to be at least 8°, it is impossible for a ground ball to be a barrel. Hence, those who are able to create weak contact by staying below the sweet-spot of the bat will be ultra-successful in limiting barrels. 

Variables that are most indicative of expected barrel rate: 

  1. Release Height Deviance from Median 
  1. Velocity 
  1. Horizontal Break Deviance from Median 
  1. Horizontal Break 
  1. Vertical Approach Angle 

One might be unsurprised that velocity and vertical approach angle find themselves atop the significance leaderboard once again, but they are actually significant in the opposite direction. The model actually argues that lower velocities are more conducive to low barrel rates. While one may be shocked by this result, there are two reasonable explanations that come to mind.  

First, barrel rate is on a per-batted-ball basis. A data analyst with minimal hitting skills such as myself has a much better chance of putting a 70mph fastball in play compared to Ben Joyce’s heater, but barrel rate is simply a measure of how well a ball was hit, given that there was a batted ball. High pitch velocities assist in juicing the exit velocity of the ball. In the MLB Home Run Derby, hitters want their pitchers to throw pitches that are soft enough to extinguish the probability of whiffing, but hard enough to assist them in hitting the ball hard. If the pitchers were to throw 30mph lobs, it would be a lot more difficult for the hitters to hit the ball as hard as they do. It is important to note that the examples provided in this paragraph are extreme. The distribution of fastball velocity at the NCAA DI level is a lot denser than 70mph and Joyce’s 105mph fastball: the 1st and 3rd quartile are separated by 4mph. However, the principle still holds true. 

The second and most significant reason is that sinkers, on average, are thrown slower than fastballs. If Pitcher X were to throw both a four-seam and a sinker, it is possible that there would be little discernible difference in the velocities of each pitch, but sinker-throwers often rely on velocity less than four-seam-throwers. As we will see shortly, sinkers are contact management kings. The average four-seam fastball velocity was 1.5mph higher than the average sinker velocity at the Division I level. 1.5mph may not seem like a ton, but it is significant enough to produce this result in the model.  

It is time to address the most significant variable in this model: release height deviance from median. The plot below will perfectly illustrate the value of release height uniqueness in contact management. 

Yakkertech, fastballs, sinker velocity, pitch release height, MLB

One essential element of inducing weak contact is releasing the ball from a non-generic release height. It appears that those with 7-foot and above release points and those with sub-4-foot release points thrive in the contact management game. Hitters feast on generic release heights. Expected barrel rate does not seem to be flattering to extreme release sides, but this is because it is physically challenging to have an extreme release height and release side. As shown in the plot, those with extreme release sides have generic release heights, and vice versa. All this tells us is that it is more important to deviate from the standard release height compared to release side.  

Yakkertech, Fastballs, MLB pitch velocity

It is time for the sinker ballers to shine. Want a fastball that limits hard contact? Kill the rise. Throwing a fastball with fewer than five inches of induced vertical break is the most foolproof way to induce weak contact. While they may not generate whiffs at a remarkable rate, they take an alternative approach to recording outs. The value of horizontal break depends on the induced vertical break of the fastball. A ton of horizontal break on a fastball that fails to sink a lot gets hit hard. A ton of horizontal break on a fastball that succeeds in generating sink does not get hit hard. Those in the 10 inches of rise, 15 inches of run area may want to experiment with a different seam orientation. If currently using a four-seam grip, they evidently struggle to generate backspin, so they should likely switch to a two-seam grip. If currently using a two-seam grip, their attempt at a sinker is not very effective, so they may be better off trying to backspin a four-seam. 

One type of fastball that we have not heavily addressed so far is the low spin efficiency fastball. I recently published a piece on Jakob Junis who saw a dramatic rise in performance after ditching his inefficient four-seamer for an effective sinker, so I will hyperlink that for a primer on inefficient fastballs. This type usually sees around 10-12 inches of rise and approximately 5 inches of run. So far, low efficiency fastballs have proven to be inadequate inducers of in-zone whiffs, out-of-zone swings, and called strikes, but they seem to be a bit more effective as weak contact generators. A good example of this at the Major League level is Max Fried’s fastball. Characteristically, Fried’s fastball has a 74% spin efficiency with 12 inches of rise and only 2 inches of run. The pitch has an unimpressive 15% whiff rate and only 22% of his fastballs result in a called strike or whiff. However, the pitch thrives in the contact management department: 5% Barrel%, 47% Ground Ball%, and an average exit velocity of 86mph.  

In theory, this version of a fastball should perform similarly to a sinker. The only difference is that true sinkers create sink by tilting the spin direction, whereas low spin efficiency fastballs limit rise by inducing a relatively high degree of gyro spin. True sinkers are generally more effective as they usually record about 10 more inches of run, but both versions should suffice in clipping the bottom part of the bat to provoke suboptimal launch angles. Low spin efficiency four-seamers often occur when the pitcher struggles to stay behind the baseball. From a biomechanical perspective, these pitchers are often more supination-friendly than pronation-friendly. We often see these pitchers thrive on the east-to-west movement plane as opposed to north-to-south. Usually those who throw inefficient four-seamers will simultaneously throw a sinker; they typically throw the former against opposite-handed hitters and the latter against same-handed hitters.  

It is easy to understand why a contact management model favors those with extreme vertical approach angles in either direction. Flat fastballs will cause the hitter to think that the pitch will drop more than it does, so they will induce extremely high launch angles that result in pop ups. Steep fastballs will indeed drop more than hitters expect, so they will induce extremely low launch angles that result in ground balls. The worst zone to be in is the middle zone, that will be susceptible to line drives and barreled fly balls. The easiest way to generate batted balls in the hitter’s sweet-spot range (8° to 32°) is by throwing a fastball with a generic shape that performs exactly as the hitter expects it to. 

Summary

The best way to summarize the conclusions drawn from this piece is by revisiting the core four purposes of a pitch and detailing which stuff traits are most significant in each domain.  

It is safe to say that the most effective way to generate in-zone whiffs on a fastball is by flattening the vertical approach angle. Vertical approach angle is a product of velocity, induced vertical break, and release height, so hard throwers with elite backspin abilities out of a low slot dominate this area of pitching. Ben Joyce is the quintessential example of a strong in-zone whiff generator. Joyce throws historically hard from a sub-5-foot release height, which creates a remarkably flat vertical angle as the pitch approaches home plate. Sinkers and four-seamers with low spin efficiency do not fare well in missing bats in the strike zone; even generic fastball shapes perform better here. As a general rule of thumb, the less the ball drops from the pitcher’s hand to the catcher’s glove, the better the fastball is at inducing in-zone whiffs. 

If a pitcher wants to throw a fastball that deceives hitters into swinging when it’s thrown outside the zone, the best advice I can give is to not throw a generic fastball. Any other version can have success in this domain if utilized correctly. Flat VAA fastballs that dominate the in-zone whiff category will often generate a ton of chases above the strike zone. Steeper sinkers will generate a ton of chases below the strike zone. To trick a hitter into swinging at a pitch they shouldn’t swing at, the pitch must end up in a different location than the hitter expects it to. The worst way to do this is by throwing a fastball that performs exactly how hitters expect fastballs to. This is also an area where pitchers with extreme horizontal approach angles succeed. Adding a level of uncertainty as to where the ball will end up on the horizontal plane only benefits the pitcher. 

In full disclosure, called strike generation is likely more of a product of command than it is stuff, but stuff certainly plays a role as well. The fastballs that best generate called strikes are the ones with extremely flat VAAs. A general theme to understand is that outlier VAAs—particularly super flat VAAs—end up in a different location than the hitter expects. Flat VAAs are also frequently partnered with high velocity. High velocity limits the hitter’s time to make a swing decision, which decreases their probability of making a correct swing decision. Outlier HAAs are beneficial as well, for the same reason that they help induce chases. Both chase and called strike generation require the hitter to make an incorrect swing decision, and the best way to do so is to throw a fastball with unorthodox angles. Because of this, it is once again important to avoid throwing a generic fastball at all costs.  

Although the in-zone whiff, out-of-zone swing, and called strike categories share a ton of overlap, contact management is an area that favors different elements of stuff. This is where sinkers thrive. The general theme that prevails is that strong contact-managing fastballs kill rise. Not only do traditional sinkers prevent barrels, but low-spin efficiency fastballs also kill rise and find the bottom of the bat. Release point is another huge factor in fastball contact management. Those with extremely high or extremely low release points succeed in staying above or sinking below the bat, respectively. Extreme launch angles in either direction only benefit the pitcher. Those in the middle-ground of the release height spectrum find themselves on the wrong side of barreled balls. To put it concisely, fastballs with fewer than 10 inches of induced vertical break and fastballs thrown from unconventional release heights limit barrels. Fastballs with generic movement profiles and generic release points do not. 

As I stated at the forefront of this piece, Stuff+ models are immensely valuable tools for pitch design and for understanding the general quality of a given pitch. However, I wanted to conduct a more detailed analysis of the relationship between fastball stuff characteristics and fastball utility. Fastball X is good or Fastball X is bad only gets a man so far. Once the games begin, it is far more useful to have a strong understanding of where a pitch thrives and where a pitch falls short. If there is one tidbit of information that you draw from this piece, let it be this: Avoid. Generic. Fastballs. 

Stuff Glossary 

Velocity: The speed of the baseball at ball release, measured in miles per hour. 

Induced Vertical Break: The distance that the ball moves vertically as a result of the Magnus spin imparted on the baseball, relative to a baseball thrown with no Magnus spin, measured in inches. 

Horizontal Break: The distance that the ball moves horizontally as a result of the Magnus spin imparted on the baseball, relative to a baseball thrown with no Magnus spin, measured in inches. 

Vertical Approach Angle: The vertical angle of the baseball as it crosses home plate, colloquially referred to as the steepness or flatness of the pitch, measured in degrees. 

Horizontal Approach Angle: The horizontal angle of the baseball as it crosses home plate, measured in degrees. 

Release Height: The vertical distance between the point at which the ball is released and home plate, measured in feet. 

Release Side: The horizontal distance between the point at which the ball is released and the center of the rubber, measured in feet. 

Release Extension: The distance between the point at which the pitcher releases the ball and the rubber, measured in feet. 

Spin Efficiency: The percentage of the spin imparted on the baseball that is Magnus spin (the spin that induces movement), as opposed to gyroscopic spin. 

One thought on “Avoid the Dead Zone: An Extensive Analysis of the Relationship between Fastball Stuff Characteristics and Utility through Four Logistic Regression Models

Add yours

Leave a Reply

Powered by WordPress.com.

Up ↑

%d bloggers like this: