The Dilemma

I believe that we have been looking at pitch quality all wrong. Many mathematical models designed to evaluate a pitcher’s stuff have been misdirected. These heavy claims are baseless without evidence, but in this piece, I plan on exploring a massive shortcoming with traditional stuff analysis, and how we can change that. We often see pitchers with A-Grade stuff on paper who struggle to translate that to strong in-game results. Our knee-jerk response is to blame their command. While this may be a valid explanation for some pitchers, it is rarely the main reason for the discrepancy between stuff and performance. Instead, the main reason revolves around a shortcoming with traditional stuff analysis.

Ever since the dawn of the pitch tracking era, pitchers have sought higher spin rates in order to induce more movement on their pitches. After all, why wouldn’t they? We’ve all seen these nasty breaking balls spin north of 3000 RPM. The hitter expects the pitch to land over the heart of the plate before it takes a sharp turn in a different direction. MLB pitchers have learned to manipulate the baseball in ways that previously could have only been achieved with a wiffle ball.  

To exemplify my point about wiffle-ball-esque pitches, I will point to a pitcher whose stuff on paper appears to be fresh out of a video game. Jordan Hicks of the St. Louis Cardinals might possess the single-most nasty two-pitch mix in MLB history. He is known for having touched 105mph on his sinker. That sinker runs 16 inches, and his slider sweeps 12 inches. A 28-inch horizontal movement differential combined with historical velocity must make Hicks unhittable. Right? 

Somehow, the results don’t seem to agree. This is Hicks’ second straight season with an ERA above 5.00. He only strikes out 23% of hitters while walking 16%. Ah! There it is! Hicks must be another example of a flame-thrower with generational stuff who is plagued by an inability to throw strikes. Except… he’s not. This season, Hicks has thrown 42% of his pitches in the strike zone. Out of 587 pitchers who have thrown at least 10 innings in 2022, Hicks ranks 270^th in zone rate. His ability to throw strikes is quite literally average. How does a pitcher with an average zone rate have a walk rate that ranks in the 1^st percentile? The answer is simple. Jordan Hicks cannot generate strikes on pitches out of the strike zone.  

You may think that the last sentence is hyperbolic, but it is not. Only three pitchers in Major League Baseball have lower chase rates than Hicks. How is that possible? Hicks throws gas with a ton of movement. How does he not induce chases?  

If a batter were to swing at a randomly-generated Jordan Hicks slider, there is a 50% chance that the batter whiffs. The issue is that batters don’t swing very often at Jordan Hicks sliders. Specifically, they swing at a mere 38% of them, and that figure plummets to 27% on sliders out of the zone. It is quite difficult to generate a strike on a pitch out of the strike zone if hitters don’t swing at it. Hicks’ issue is rather clear and far from uncommon. His pitch mix generates too much movement. 

Yes, you read that correctly. Hicks’ 28-inch horizontal break differential between his two pitches is counterproductive. To understand why this is, we must first understand ball flight. There comes a point in a pitch’s journey from the pitcher’s hand to the catcher’s glove where the hitter must make his swing decision. This is a product of pitch velocity, pitch location, bat speed, hitter reaction time, among other factors. When Hicks throws a slider with 12 inches of sweep, the pitch likely begins to sweep earlier than a slider with 4 inches of sweep. While there is such a thing as “late break” and it can be quantified through measures of seam-shifted wake, I am only referring to movement induced by Magnus spin.  

The issue is that Hicks is unable to form an effective tunnel between his sinker and slider. The two pitches diverge from their tunnel earlier than pitches with smaller movement differentials, and based on Hicks’ abhorrent chase rates, I speculate that the two pitches become detectable before the hitter makes his swing decision. The moment a hitter sees arm-side run, they know it is a sinker. The moment a hitter sees glove-side sweep, they know it is a slider. Given the large magnitude of each, the hitter can detect the pitch type early on in the pitch’s flight.  

I use Hicks as an example, but he is far from the only pitcher with this issue. Within the baseball analytics community, there seems to be an infatuation towards sweepers, such as the one Hicks possesses. Sweepers are slider-variants with minimal spin-induced break on the vertical plane, and a ton of glove-side sweep. Hicks’ 12-inch sweeper certainly falls in that category, but we’ve seen sweepers touch the 20-inch mark.

Many pitchers waltz into pitching labs and attempt to garner as much sweep as possible on their slider, thinking they’re developing a nasty pitch. But is this truly beneficial? Does the pitch fit within the rest of the pitcher’s arsenal? I don’t want to leave you with the impression that sweepers are bad per se, but I will confidently state that they are pitches that should be thrown by a select number of pitchers. 

A strong example of a pitcher who excels with a sweeper is none other than Shohei Ohtani. Ohtani’s sweeper averages 14 inches of sweep, which may seem like too much based on the Jordan Hicks example, but it is not. 45% of Ohtani’s sweepers result in a called strike or a whiff. That statistic lies at 23% for Hicks. 50% of Ohtani’s sweepers generate a swing, compared to 37% for Hicks. One might naturally assume that the reason Ohtani’s sweeper is more effective is that he throws more than two pitches, but that is not the main reason. Ohtani actually throws his sweeper 6% more than Hicks does. Technically speaking, Ohtani’s sweeper is his primary pitch.  

The reason Shohei is able to succeed with a sweeper is because of his fastball. Ohtani throws a low spin efficiency four-seamer than only generates three-inches of run. Despite generating more sweep on his slider, his fastball-slider horizontal movement differential is only 17 inches—11 fewer than Hicks. The two pitches form a tunnel that is far more effective than Hicks’ sinker-slider combination. 

The Truth About Breaking Balls

I recently published an article on the relationship between fastball stuff characteristics and fastball utility. While I was conducting a similar analysis for breaking balls, I discovered a rather telling conclusion.

This plot is based on 2022 NCAA Division I Yakkertech data. It illustrates a logistic regression model I made that outputs the expected chase rate on a breaking ball based on stuff characteristics. It is important to note that each of these breaking balls is treated as if they exist in a vacuum. They are not relative to the rest of the pitcher’s arsenal. Any negative horizontal movement represents glove-side sweep, and any negative induced vertical break represents spin-induced drop.  

It is glaringly evident that breaking balls with a ton of sweep and drop are rarely effective as chase pitches. Some mega sweepers can be effective, but this only works effectively if the pitcher’s fastball does not run a lot. Ohtani is an example of a pitcher with an effective sweeper, but he possesses a rare fastball that allows it to work. I believe there are six types of breaking balls: cutters, slutters, gyro sliders, sweepers, slurves, and 12-6 curves. The two that appear to be the most effective are the cutters and the gyro sliders. Interestingly, these are the two variants that often see the least spin-induced movement.  

I am not saying that there is no purpose to throwing breakers with a ton of spin-induced movement. Quite the contrary. There is a rather valuable purpose. 

This plot appears to be the polar opposite of the prior one. It represents the expected likelihood of inducing a swing on a middle-middle breaking ball. The lower the probability, the more effective the pitch as a called strike generator. Loopy breakers with double-digit inches of drop and sweep can reach expected middle-middle swing rates as low as 40%. So, please do not leave with the belief that loopy breakers are bad. They may be ineffective at generating chases, but they are remarkably effective called strike pitches. Throw them in the zone, not out of it.  

I’ve focused on horizontal differentials thus far, but I would argue that vertical differentials are more indicative of tunneling prowess. To analyze vertical movement in this article, I will be citing long-form vertical movement instead of the traditional short-form induced vertical break that is shown in the plots above. I chose to use long-form movement because it represents the magnitude of drop that a pitch exhibits as a result of spin and velocity. I find this metric more useful when referencing vertical movement differentials because it incorporates spin-induced drop and velocity-based drop.  

While too much separation between two pitches is evidently detrimental, the opposite exists as well. Padres Lefty Sean Manaea’s two most-used pitches are his sinker and his changeup. Unfortunately for Manaea, his changeup has been the 5^th worst changeup in MLB this season per run value. After glancing at its movement profile, I can’t say I’m surprised. The pitch drops only 11 inches more than his sinker, while running 0.2 inches less. This is not enough separation! Even if it takes the hitter longer to visually differentiate the two pitches, they are able to guess wrong and still be successful. Manaea’s changeup woes are not uncommon for sinker-ballers. It is easier to separate a four-seam and changeup than a sinker and changeup. 

We have established that too large of a movement differential is ineffective, and too little of a movement differential is ineffective, so what is effective? There exists a sweet-spot differential that catalyzes an effective tunnel while still differentiating the movement profile. Now we have reached the hard part: finding that sweet-spot.

The Vertical Movement Sweet-Spot

To determine the ideal vertical movement differential, I looked at drop differentials between a pitcher’s primary fastball and their breaking ball(s). I opted to use chase rate as the metric that would benefit the most from a strong tunnel. It seemed appropriate, as chase rate measures the ability to induce a wrong swing decision.

I soon realized that breaking balls and off-speed pitches would need to be separated, as they possess different, albeit similar sweet-spots. For breaking balls, I limited the sample to breakers with fewer than 17 inches of horizontal break differential off of the fastball, as to somewhat control for extreme horizontal breakers. Among these breaking balls, here are the conclusions I drew: 

1. The vertical drop differential sweet-spot is between 16 and 26 inches. 

2. Among breaking balls with top quartile chase rates, 54% fall in this sweet-spot. 

3. Among breaking balls with bottom quartile chase rates, 36% fall in this sweet-spot. 

4. The average chase rate of breaking balls that fall in this sweet-spot is 33%. 

5. The average chase rate of breaking balls that do not fall in this sweet-spot is 30%. 

These may not seem like extreme differences, but they are significant. a 30% chase rate falls in the 59^th percentile. A 33% chase rate falls in the 85^th percentile. The value of throwing a breaker in that sweet-spot is tremendous if the pitch is to be used as a chase pitch. It is important to remember that this sweet-spot is not a hard-gated fence. Breakers with 14 inches of drop differential should not be treated the same as breakers with 40 inches of drop differential. Edwin Diaz has one of the most effective sliders in Major League Baseball, and he only records 14 inches of drop differential. It is very possible to have an effective breaking ball that does not fall within this range. However, this is the range that appears to create the most effective tunnels with one’s primary fastball. 

Here are the ten breaking balls with the highest whiff-per-pitch rates, with their vertical drop separation from their primary fastball in parenthesis. 

1. Jacob deGrom SL (14.9) 

2. Andrew Chafin SL (19.6) 

3. Edwin Diaz SL (14.2) 

4. Dylan Lee CU (22.8) 

5. Ryan Pressly SL (18.3) 

6. Liam Hendriks SL(20.5) 

7. Dominic Leone SL (26.0) 

8. Andres Muñoz SL (16.1) 

9. Tanner Rainey SL (22.5) 

10. Ryan Helsley SL (22.7) 

Every one of these breaking balls falls in the sweet-spot except for deGrom’s and Diaz’s, and even those two aren’t too far out of it. deGrom and Diaz also excel on the horizontal break plane as well, which we will dive into shortly. Within the range, it appears that the closer one can come to a 20-inch differential, the better off they are. 

The plot below demonstrates that a drop differential between 10 and 30 inches gives one a strong chance at inducing chases. Any smaller than 10 and any larger than 30 makes it a lot more difficult. I chose 16-26 inches to be the sweet-spot, because this is where we see the majority of elite breaking balls.  

To prove the importance of creating a fastball-breaker drop differential in this range, I will point to a Major League example. In fact, I will point to a starting pitcher without an elite fastball who instantly improved with a slight modification to his slider. Angels Starter Reid Detmers throws a fastball that is not necessarily bad. I think average might be a better word to describe it. Prior to his demotion to AAA in late June, he had yet to see sustained Major League success. Up to that point, the only consistency that his slider had shown lied in the fact that it was consistently different every start, as I tweeted in July.  

https://twitter.com/MaxwellResnick/status/1553787007280861184

Detmers’ fastball drops 14 inches. Based on what we have learned so far, he should search for a slider with between 30-40 inches of drop. The plot below shows Detmers’ vertical drop on his slider in each start this season. I placed a dotted blue line at the 34-inch benchmark, which represents 20 inches of vertical drop differential from his fastball.

A proper fastball-slider tunnel was non-existent early in Detmers’ season. Once June hit, he began to discover something stronger, but it still lacked consistency. Now, in August, Detmers has found one of the most lethal secondary offerings in the game of baseball. His slider chase rate in the month of August lies at 50%. For reference, it was 17% in April. One might think that the succeeding sentence is an exaggeration, but I truly believe it. Detmers’ ability to find a slider that creates a proper tunnel with his fastball turned him from a back-of-the-rotation guy to an elite starting pitcher.  

Unfortunately, not everyone has found the same success as Detmers. White Sox Righty Michael Kopech has long been considered one of the more electric players in the game of baseball, mostly due to his elite fastball. At the Major-League level, his fastball has performed exactly how everyone expected. However, he has yet to burst into the upper-echelon of pitchers due to his inability to develop a strong secondary. His slider is a prime example of a pitch with a differential that is too large. He generates 29 inches of drop differential between his fastball and slider, which is outside of the sweet-spot range, and a far cry from the ideal mark of 20 inches. This same slider records a whiff under 10% of the time. Why is that?  

As we discussed earlier, the two pitch types likely become detectable before the hitter is forced to make his swing decision. I call this a short tunnel. Even if the pitches start on the same plane, it does not take long for them to break off. Kopech could realistically burst into the elite tier of pitchers if he can develop a harder breaking ball that drops less. A slight modification could be career-altering for him. He is not alone in that category. There are many other elite fastball-throwers whose lack of secondary offerings hold them back from reaching their full potential.  

I ran a similar procedure for fastball-changeup combinations (splitters are grouped in with changeups), and here are the conclusions I drew: 

1. The vertical drop differential sweet-spot does not really exist. As a general principle, more drop separation between fastball and changeup bodes well for pitchers. However, the number that appears to be the pre-requisite for changeup success is 14. 

2. Among off-speed pitches with top quartile chase rates, 57% have more than 14 inches of vertical separation. 

3. Among off-speed pitches with bottom quartile chase rates, 48% have more than 14 inches of vertical separation. 

4. The average chase rate of off-speed pitches with more than 14 inches of separation is 38.3%. 

5. The average chase rate of off-speed pitches with fewer than 14 inches of separation is 35.8%. 

Admittedly, these results are less extreme than that of breaking balls. However, they are certainly not insignificant. The reason there is less extremity is due to the smaller range of drop differentials among off-speeds compared to breakers. The standard deviation of fastball-changeup vertical drop differential is 4.7. The standard deviation of fastball-breaker vertical drop differential is 9.9. This makes sense in theory. As I stated earlier, there are six breaking ball variants. The only two off-speed variants are changeups and splitters, and even those two usually perform similarly. 

Although the data did not show an upper limit to ideal vertical drop separation with off-speeds, one likely exists. Pitchers have yet to reach it. I imagine that if a pitcher got 50 inches of separation between their fastball and changeup, the two pitches would not tunnel very well. However, the realistic maximum we have seen is around 30 inches, and that differential appears to be effective.  

Earlier in this piece, I referenced Sean Manaea as an example of a pitcher who struggles to separate his fastball and changeup. This is a very common issue for those who throw sinkers as their primary fastball. Sinkers possess short-form movement profiles similar to changeups, so the only real separator is velocity. For this reason, I would advise sinker-ballers to explore slower changeups with fastball-like arm speed, Manaea included.  

I want to talk about one of Manaea’s teammates as an example of this velocity differential. Padres Righty Nick Martinez has a rather standard blueprint for pitching. Sinker-cutter combination to righties, four-seam-changeup combination to lefties, and a loopy curveball to mix in once and while. Oftentimes, pitchers with this blueprint throw poor four-seams, but they fear that if they nix the four-seam, their changeup success will plummet due to a killed tunnel. Martinez is no different. Every pitch of his has performed at an above-average level this season, except for his four-seam, which possesses an abysmal run value of 8. He must throw it to avoid killing the changeup, though. Right? 

Martinez’s changeup is a unicorn pitch. He throws it 14mph slower than his four-seamer, and imparts 13 fewer inches of spin-induced break on it. His changeup drops 25 inches more than his four-seamer. This exceptional vertical separation is the main reason his changeup is so successful, so limiting four-seam usage would likely have a negative effect on the changeup. But the 21 inches of drop differential between Martinez’s sinker and changeup is still very strong. I am not recommending that Martinez ditches the four-seam altogether, but he has less reason to be fearful of it than many similar pitchers, due to the large velocity differentials between his two fastballs and his changeup. 

Before moving on, I would like to clarify that spin-induced drop should always be preferred over velocity-induced drop. Although they may produce similar pitch shapes, velocity is the key determinant of a hitter’s reaction time to swing. Creating large separation without over-extending a hitter’s time to make his swing decision is essential. Yankees Reliever Ron Marinaccio owns one of the most potent changeups in the game. His vertical drop differential from his fastball is nearly 24 inches, which far surpasses the 14-inch benchmark. He achieves this massive separation with a 15-inch induced vertical break differential that few pitchers are able to achieve.

The Horizontal Movement Sweet-Spot

Horizontal break is certainly less complicated than vertical break, due to the absence of gravity. As we all know from existing on Earth, gravity does not operate on the horizontal plane, so there is no variation in short-form and long-form horizontal break.  

As I did in the vertical drop section, I decided to look at the breaking balls in Major League Baseball with the highest swinging-strike rates. I analyzed the horizontal movement differential between said breaking ball and that pitcher’s primary fastball. The results were jarring. Before sharing them, I want to establish that the width of the strike zone is 17 inches. 

1. Not a single breaking ball in the top 20 of swinging-strike rate had more than 17 inches of horizontal separation off of its primary fastball. 

2. Not a single breaking ball in the top 35 of swinging-strike rate had more than 20 inches of horizontal separation off of its primary fastball. 

3. Among those top 35, 10 breaking balls had fewer than 10 inches of horizontal separation off of its primary fastball. 

4. Among those top 35, 21 breaking balls fell between 10-17 inches of horizontal separation off of its primary fastball. 

Remember the opening of this piece when I analyzed the Jordan Hicks dilemma? Well, Jordan Hicks is not alone. The breaking balls that are most successful at inducing swinging strikes do not distance themselves from their fastball by more than 20 inches. Even 20 inches is flirting with the devil. I would deem the sweet-spot to be within 6-16 inches. Believe it or not, I would rather be within 6 inches of separation than beyond 16 inches of separation, unless the pitcher also throws an intermediate pitch to bridge the gap. 

Once again, do not walk away from this article believing that sweepers have no value. Sweepers approach the plate at a very sharp horizontal angle, making them difficult to barrel. Hitters struggle to gauge where a sweeper will land horizontally. However, if a mega-sweeper is paired with a running fastball, hitters will lay off the breaker. Hicks is not alone. 

I want to provide a counterexample to Hicks. So far, we have established that an ideal fastball-breaker tunnel is formed with 16-26 inches of vertical drop separation and 6-16 horizontal break separation. Braves Starter Charlie Morton has a five-pitch arsenal, but he throws two of those pitches way more than the other three: his fastball and curveball. On the surface, the two should not tunnel well. They are separated by 34 inches of vertical drop and 26 inches of horizontal break—both of which are way beyond their respective sweet-spots. Nonetheless, Morton’s curveball possesses a 36% chase rate, which is comfortably above average for breaking balls. 

How does he do this? Although Morton primarily relies on two pitches, he also throws a cutter 10% of the time.  

Pitch Type	Usage	Vertical Drop	Horizontal Break
Fastball	34%	18	10
Cutter	10%	32	-2
Curveball	38%	53	-16

Although the fastball and curveball do not create an ideal tunnel, the cutter bridges the gap. The fastball and cutter create a strong tunnel, and the cutter and curveball create a strong tunnel. As a result, the presence of the cutter helps bridge the gap between Morton’s two primary offerings. My best advice to a hitter facing Morton would be to pretend the cutter doesn’t exist. This may sound facetious, but it is not. You are better off fixating on the fastball-curveball combination that does not tunnel well, and if he beats you with the cutter, you may tip your cap on your long walk back to the dugout.  

In the case of Reid Detmers, his path to developing a strong tunnel revolved around slider renovation. He took a pitch in his arsenal and crafted it to fit his fastball. Morton could have done the same, but his curveball has been his most lethal weapon for his entire career. Instead, he developed a cutter that bridged the gap between his fastball and curveball. This is another feasible and effective method of engineering an effective tunnel. 

What conclusions can we draw from my exploration of horizontal break separation between fastballs and breaking balls? Believe it or not, I see no evidence that more separation is better. I should establish that a breaking ball should not have the exact same movement profile as a four-seam fastball. However, this is not an issue that Major League pitchers face. As long as there is approximately 6 inches of horizontal separation, I see little evidence that suggests that more is better.  

Breaking balls that induce swinging strikes appear to do so by dropping below the bat, instead of trying to miss the end of the bat. This makes sense in theory, considering that bats are far longer than they are wide. It is a lot easier to induce a straight whiff by missing over or under the bat than by surpassing the length of the bat. For this reason, I would always recommend prioritizing vertical tunnels over horizontal tunnels, at least for the purpose of inducing whiffs. Does the same hold true for changeups? 

It does. There is not a strong relationship between fastball-changeup horizontal separation. Many pitchers see success with running changeups, such as Brandon Woodruff and Shane McClanahan. Others see success with changeups that barely run more than their fastball. Notable sinker-baller Logan Webb throws a changeup with half as much run as his sinker.  

The best way to view fastball-changeup separation is through a coordinate plane. If a pitcher were to plot their fastball on a long-form movement coordinate plane, they should strive to, within reason, maximize the Pythagorean distance between their fastball and changeup. The pitcher should first prioritize vertical separation. Once they feel as if they’ve maximized the drop differential, any additional run is a solid bonus. 

Although I have claimed that the pitcher’s goal should be to maximize fastball-changeup separation, an upper limit does exist. Simply put, pitchers have yet to scrape this upper limit. We see fastball-slider combinations with 35 inches of horizontal break differential. We do not see anywhere near this magnitude with fastballs and changeups. Horizontal separation that reaches double digits is rare. I imagine that if pitchers were to achieve 30+ inches of horizontal separation between their fastball and changeup, it would be quite detrimental. However, within the limits of our current version of baseball, I have yet to see a pronounced upper limit. 

I’ve spent the duration of this piece solely referring to a pitch’s ability to generate chases and miss bats. I have done this because I wholeheartedly believe that the ability to induce a chase and a whiff is the most valuable attribute a pitch can have. Nonetheless, there are other methods of recording outs. I theorize that horizontal separation could play a role in inducing weak contact.  

Anyone who has ever held a baseball bat knows that the barrel of the bat produces the hardest contact. That might explain why there is a commonly-cited quality of contact metric titled “Barrel%”. The endcap of the bat and the handle of the bat do not produce favorable contact. Pitchers who operate predominantly on the horizontal movement plane likely attempt to reach these extremes of the bat, instead of trying to miss it altogether. The issue, as we saw with Hicks, is that too much movement will simply reduce the likelihood of a swing. It is rather difficult to induce weak contact if the bat never leaves the hitter’s shoulder. 

I also want to talk about the most lethal pitch type on the planet: the splitter. Splitters are dominant, and if they were easy to throw, everyone would do it. Those who do throw them often see remarkable success with them. Despite the infrequency of the pitch, 5 out of the top 8 off-speed pitches in swinging-strike rate are splitters. The reason I opted to bring up splitters in this section is because splitters rarely garner much horizontal separation from their corresponding fastball. In Yu Darvish’s case, his splitter runs less than his fastball (which isn’t too uncommon).  

Analysis & My Thoughts

To fully analyze the effects of separating one’s fastball and breaking ball, I decided to create a very simple metric called “Tunneling Score”. Essentially, it sets the ideal separation as 20 inches of drop differential and 10 inches of horizontal differential, and measures how much a breaking ball strays from this ideality. The table below shows the 2022 leaders in fastball-breaker tunneling score, along with the chase rate of their breaking ball.

Pitcher & Pitch	Vertical Separation	Horizontal Separation	Chase Rate
German Marquez CU	20.2	10.0	39.0%
Shane Bieber SL	20.0	10.9	41.2%
Patrick Sandoval SL	20.3	10.7	42.1%
Kyle Nelson SL	21.1	10.0	36.5%
Blake Snell SL	18.7	9.6	40.8%
Matt Foster SL	19.5	11.2	30.0%
Scott Barlow SL	19.9	11.7	35.7%
Taylor Clarke SL	19.2	9.0	33.8%
Chris Stratton SL	18.2	9.9	39.2%
Austin Gomber SL	19.5	11.4	30.6%

Every breaking ball in the top 10 of tunneling score owns a chase rate of at least 30%. The correlation between tunneling score and chase rate sits at 0.3. While this figure isn’t too extreme, I do believe that there is certainly a relationship between the two variables. There will always be exceptions, but I firmly believe that pitchers should strive to create that 20-10 differential if they intend to use their breaking ball as a chase pitch.

I decided to calculate the distance between a pitcher’s fastball and changeup on a long-form graph through the Pythagorean formula. Although the relationship was not overwhelming it was existent. The changeups that were more separated from their corresponding fastball performed better. This supports everything we have discussed so far regarding the relationship between fastballs and changeups.

I have dedicated the entirety of this piece to the relationship between fastballs and secondaries, but the truth is that secondaries can play off of other secondaries. There are several examples of pitchers who throw a slider and a curveball, but some of the elite starting pitchers utilize three glove-side movers. Gerrit Cole is a perfect example of this.

The black lines that connect datapoint are used to represent an effective tunnel. His fastball and slider are separated by a 24 inches of drop and 15 inches of horizontal break, which catalyzes an in-zone fastball out-of-zone slider tunnel. His cutter and curve are separated by approximately 28 inches of drop. This is outside of the sweet-spot range that we previously established, but given the three levels of glove-side breakers, he is still able to use his curve as an effective pitch in and out of the zone. Shane Bieber is another ace who possesses a cutter, slider, and curveball. The ability to create tunnels between non-fastball offerings is all-the-more beneficial for the pitcher.

In full disclosure, I used to possess the belief that pitchers with bad fastballs should ditch that bad fastball altogether. While this may be a fair solution for some pitchers, I neglected a key element of pitching in my prior analysis. That bad fastball often operates as a set-up pitch for a pitcher’s secondaries, and abandoning the fastball would likely have a negative effect on the rest of the arsenal.

A good example of this is Patrick Sandoval of the Angels. Sandoval’s four-seam fastball is ineffective, regardless of how you want to look at it. The velocity is average at best, the shape is generic, it gets hit hard, and it generates an abysmal number of whiffs. He also throws a sinker, which doesn’t have great characteristics either, but it is more productive than the four-seamer.

Sandoval makes his money with his slider and changeup. Both pitches generate a called strike or whiff above 30% of the time. How are they so successful? They tunnel perfectly off of the four-seamer. Maybe perfect is a stretch, but it isn’t far from it. Sandoval’s fastball and slider are separated by 20 inches vertically and 11 inches horizontally. Sandoval’s fastball and changeup are separated by 16 inches vertically and 10 inches horizontally. If he were to ditch the four-seam in favor of the sinker, he would ruin the elite tunneling effect he currently has in place.

Sandoval’s upside will always be capped by the presence of a bad fastball, but he does a good job limiting that presence. His four-seamer is actually his tertiary pitch. He throws it 24% of the time, which is enough to create these elite tunnels, but not enough to cause irreparable damage to his production. As a general principle, pitchers in Sandoval’s shoes should strive to throw their four-seamer between 20%-30%. The recommended percentage obviously varies by circumstance, but the 20-30 range seems to work well for a lot of similar pitchers.

Conclusion

It is time to rethink pitch design. Trust me. I do not subscribe to the old-school belief that a pitcher must “establish his fastball”. I am, in fact, an advocate of throwing your best pitches more. However, with a few exceptions, I believe that pitchers should craft their breaking balls and off-speed pitches off of their fastball. 

The reason I say this is because the fastball is the least malleable pitch. Outside of slight modifications to grips, the only real variable is four-seam versus two-seam. In a single pitch design session, it is unlikely that a pitcher will be able to see a wide variance of results with his fastball. At the end of the day, the pitcher’s goal with their fastball should be to throw it as hard as they can. 

However, with sliders, changeups, and curveballs, there is a ton of potential variance. Through grip, release, and cue modifications, a pitcher has a ton of ways to craft a breaking ball to fit their fastball’s needs. If a pitcher’s slider drops too much to create an effective tunnel, instruct them to throw it harder. If a pitcher’s slider doesn’t drop enough, instruct them to impart more topspin on the ball. There isn’t a hard-and-fast answer to every problem, but there is almost always room for exploration with secondary offerings. 

Next time you look at a pitcher’s pitch data, fight the urge to automatically assume that more movement is better. In the majority of cases, it is not. Instead, ask yourself why the pitcher with 30 inches of horizontal break differential between their sinker and slider fail to generate chases or whiffs. There is almost always a reason, and that reason is almost always related to excess movement. 

One would be shocked at how many Major League pitchers fail to generate chases due to the poor tunnels their pitches form. Even pitchers who pitch for analytically progressive organizations. Cristian Javier of the Houston Astros is putting together a really solid season, but his one shortcoming is his chase rate. Operating as a two-pitch pitcher, it is crucial to maximize the effectiveness of his fastball-slider combination. 28 inches of vertical drop differential and 22 inches of horizontal separation do not cut it. His chase rate is well below average, which triggers a poor walk rate. The domino effect is real.  

I opened the article with a critique mathematical stuff models, so let me elaborate. I am unsure of the composition of every model out there, but I have noticed a common trend. Stuff+ models factor in movement differentials between pitches, but they do so in a linear fashion. They neglect the fact that ideal movement differentials often exist on a spectrum. Similar to a variable such as launch angle, too much and too little are both detrimental. As we’ve covered throughout this piece, there usually exists a sweet-spot that catalyzes ideal tunnels. Stuff+ models should incorporate these sweet-spots, as to not overvalue pitchers who generate too much movement. 

We often see pitchers overperform or underperform stuff models, and our gut reaction is to attribute the gap to command. Pitcher X has a 2.00 ERA and 50-grade stuff? He must have elite command! While that may occasionally be the case, I believe that the majority of these cases can be attributed to effective or ineffective tunneling. 

We’ve already discussed Jordan Hicks in length as a guy whose stuff is better on paper than in practice, but let’s look at an opposite example. Tony Gonsolin is putting together quite the season for the Dodgers. His stuff is respectable on the surface, but I doubt anyone is writing home about it. Many models grade Gonsolin’s stuff as below average. 

However, Gonsolin’s arsenal is far from below average. He generates 16 inches of drop differential and 12 inches of horizontal separation between his fastball and slider. Both figures fall in the ideal sweet-spot. His fastball and splitter are separated by 23 inches of drop and 4 inches of run, which is remarkable. He also sports a curveball that operates as one of the best called strike pitches in Major League Baseball. See how the conventional approach to stuff evaluation does a disservice to Gonsolin? 

To recap, here are the 10 conclusions I hope you draw from this piece: 

1. More movement is not always beneficial. 

2. To form an ideal tunnel, fastballs and breaking balls should be separated by 16-26 inches of vertical drop. 20 is the number to strive for. 

3. Any breaking ball with more than 30 inches of vertical separation from its corresponding fastball should be used more as a called strike pitch than a chase pitch. 

4. As a general principle, more vertical separation between fastballs and off-speeds is beneficial. There may be a point where vertical separation becomes counterproductive, but pitchers have not approached that extreme level yet. 

5. 14 seems to be the essential benchmark of vertical separation between fastballs and changeups. 

6. A breaking ball with more than 17 inches of horizontal separation from their corresponding fastball rarely succeed as whiff generators, since they fail to induce enough swings. 

7. Ideally, a breaking ball should possess between 6-16 inches of horizontal break differential from its corresponding fastball. 

8. Fastball-changeup horizontal separation is not as important as vertical separation, but it is still beneficial to create separation on multiple planes.

9. Splitters are lethal weapons. 

10. As a general principle, vertical drop separation is more indicative of whiff generation than horizontal separation, due to the anatomy of the bat. 

There will always be exceptions to these rules, so do not automatically dismiss an effective tunnel if it breaches one of the ten conclusions listed above. These should simply be guidelines for pitcher evaluation and, more importantly, pitcher development. 

I hope you learned something from my research. I certainly learned a ton while conducting this analysis. Thankfully, we are only scratching the surface of our understanding of ball-flight tracking data. 

I would like to give a quick shoutout to a piece written by writers at prospectslive.com on pitch tunneling. Their analysis helped assist my understanding of this woefully-misunderstood science. If you are curious to learn more about this fascinating topic, I highly recommend checking it out.