Quantifying Pitch Quality with Stuff +

As more fans, coaches, and organizations are becoming versed with pitch data, it can be very difficult to narrow down which variables are actually important to the success of a pitch. While sometimes we can intuitively say x fastball is better than y fastball, it's important to have a model that backs that claim up for when the intuition isn't as clear. Stuff + is a analytical model that allows us to view the quality of the pitch in relation to the league average. It takes different variables and weighs them based on the most important factor to the success of the pitch. It is also important to note that this model shouldn't be used exclusively to evaluate the quality of a pitch. We will see some instances later on where this is the case. With that being said, let's jump right into my Stuff + model.

While there are many things that go into the outcome of a pitch, I have narrowed my model down to 4 individual factors that are all measured from your standard pitch tracking systems (Rapsodo, Trackman, etc.)

Velocity

Vertical Break

Horizontal Break

Release Height

I personally don't like to include any Spin Rate measurements in my models. Spin Rate is extremely difficult to control as a pitcher and you can still create significant amounts of movement with an average or below average spin rate. I think horizontal and vertical movement better capture the effectiveness of a pitch, hence why I used those in my model instead of Spin Rate. There is a strong argument for the use of Spin Axis (or tilt), however with the use of Trackman data, the data we would gather would pretty much mirror the movement numbers. If Hawkeye data was publicly available, we could measure spin deviation and thus making it a necessary variable. Until that data becomes available, Spin Axis will be left off the model.

The last consideration for the model would be release extension. This is a variable that I wanted to use because I think it has some impact, but Hawkeye is the only system that measures it accurately. Because of the inaccuracy of this measurement for Trackman data capture, I chose to leave out release extension.

These 4 factors were separately weighted and adjusted for each pitch type based on what we know about each pitch. For example, high carry high velo 4-seamers were rewarded heavily from the model while high velo low carry sinkers were rewarded by the model.

Before diving into the specific pitches and numbers, I briefly want to mention how release height was taken into account for all pitches. Pitchers were not necessarily rewarded for having a high release height or a low release height compared to the average. Instead, they were rewarded if their movement profile deviated from the average movement from that arm slot. A guy could have a weird arm slot, but if his pitches move as expected, he won't be rewarded as much as a pitcher who has an average arm slot who's pitches move differently than expected.

The numbers in this model mimic all other + models such as wRC+ and OPS+. Players that receive a 100 have a league average pitch. Players that receive a 120 have a pitch that is 20% better than league average. Conversely, a player that receives an 80 is 20% worse than league average. It's also important to note that a 110 4-Seam Fastball is not necessarily equivalent to a 110 Slider. Because each pitch is weighted amongst themselves, they may differ in the score given but have the same percentile ranking.

If you haven't already read my previous blog about ideal pitch clustering, it may help to clear the air about some of the discussion in this post. In that article I discussed whether or not there were better ranges for pitches to be in, and if they did exist, how much better would those results be.

4-Seam Fastball

Gerrit Cole is the king of the 4-Seam Fastball with a Stuff + rating of 165. Obviously because it's Gerrit Cole we see every 5 days how dominant his fastball is, leaving no surprise that he comes out at #1. His Fastball gets 19 inches of carry while also getting almost 11 inches of run. He further elevates this pitch by throwing it at an average of 98 MPH. What's even more astonishing (yet not included in the model) is that Cole still has about 98% Spin Efficiency while getting 11 inches of run. This leads to a Fastball that jumps out of the hand and makes it incredibly difficult to hit.

Some other notable names in the top 25 are Liam Hendricks, Shane McClanahan, Michael Kopech, and Hunter Greene. The Fastball is one of the easier pitches to quantify. The better the velo and carry, the better the pitch will be. The trick to this metric is when two players are put together, one with superior speed and the other with superior carry. This model is able to decipher and figure out which pitcher has the better Fastball based on the 4 characteristics. This relationship can be shown between Chris Archer and Walker Buehler. Chris Archer has a Fastball of 19 inches carry at 93.1 MPH, while Walker Buehler has a Fastball of 17 inches carry at 95.2 MPH. The model indicates that Buehler has the slight edge of an 8% advantage over Archer (114 to 106).

Sinker

Even though he has been out since May with a back injury, Anthony Bender leads the MLB in Sinker Stuff + at a 176. Bender averages 2.5 inches of carry with about 17.5 inches of run on his sinker. His pitch really falls off the table in comparison to other sinkers, especially since it is thrown at 98 MPH. While there are pitchers that have more sink than Bender's Sinker, none throw it at the velocity and arm angle that Bender does, thus making his pitch superior.

Some other notable names in the top 25 are Sandy Alcantara, Clay Holmes, Edwin Diaz, and Blake Treinen. The one thing all 5 of these pitchers have in common is their elite velo, combined with good movement. While their movement may not stack up with other pitchers (like Logan Webb who has significantly more sink on his sinker) it certainly makes up for it with a boost in velocity. The ball isn't going to have nearly as much movement when it is thrown harder as it has less time in the air for drag to impact it. Because of this, the model is more forgiving with movement if it is coupled with elite velo (97+). However, if a pitcher has decent velo and decent movement, the model will likely punish the player for not exceling at one or both characteristics

Cutter

The best cutter in baseball belongs to Shawn Armstrong with a Stuff + of 150. Armstrong has more of a "slutter" movement with 5.5 inches of carry and 5.5 inches of cut. The Slutter was a term coined by Tyler Glasnow to describe his Slider movement as being a combination of a cutter and a slider. Armstrong has this movement but for his cutter. He also throws the pitch at 91.5 MPH.

Notable names in the top 25 would include Jose Alvarado, Luis Severino, Julio Urias, and Corbin Burnes. The cutter has a lot more variability as there are many ways to succeed with the pitch. Some may use it as a replacement of a slider, while others may use it to bridge the gap between a big slider and a high carry fastball. Cutters typically aren't used as a go to swing and miss pitch, and therefore show a lot less variability in values.

Slider

Emmanuel Clase wins the best slider with a Stuff + of 147. There are a lot of quality sliders, thus pointing to the much lower score. Regardless, Clase's slider averages about 6 inches of horizontal movement and 0 inches of vertical movement. However, this movement does not win the race for Clase. Instead it is his average of 92 MPH on the pitch. The speed on this pitch is ridiculous and is the main contributor to his success. In the video below, Clase throws a slider at 96 MPH (how that is even possible, who knows).

Notable names that round out the top 25 include Corbin Burnes, Gerrit Cole, Blake Treinen, and Daniel Bard. The slider is the one pitch where velocity is the big separator, as shown through these 5 players. As discussed in my pitch profiling post, there is no clear movement pattern that shows significantly more success than the rest. The only bad sliders are the ones that get too much vertical break, turning more into slurves which lose the bite of the better sliders. Because of this similarity in movement success, velocity is the biggest separator between a good slider and an elite one.

Curveball

Josh Sborz has the best curveball with a Stuff + of 147. Again, there are plenty of plus curveballs so there is not as much separation among curveballs. Sborz has a true breaker as he averages about 12 inches of vertical drop and 12 inches of horizontal movement. Sborz also throws his curveball at about 85 MPH, which is well above league average.

Some other notable names in the top 25 are Gerrit Cole, Clark Schmidt, Seth Lugo, and Ryan Pressly. The curveball is the most balanced pitch among the 6 included in this list. Joe Kelly throws the hardest average curveball at 88.6 MPH, yet only has a 116 on Stuff +. Jhoan Duran has the second hardest curveball at 87.5 MPH, yet only has a Stuff + of 106. Having the best of the best in velocity only puts a curveball slightly above average. On the other hand, Rowan Wick has the second largest vertical break at 18.5 inches, yet only has a 112 Stuff +. This shows that in order to have an elite curveball, you need to have an elite combination of movement and velocity, something that everyone at the top of the list has.

Changeup

Last on the list is the Changeup, which for the convenience of the metric has been combined with the splitter as both pitches serve the same purpose. It is no surprise to see Devin Williams at the top with a Stuff + of 176, about 11% higher than second place changeup. Williams' changeup is known as an "airbender", getting -3 inches of vertical break and 17 inches of run. While this doesn't sound super impressive, it is really put into perspective when you look at his fastball, which averages 17 inches of carry. He creates 20 inches of separation from his fastball at an average of 84 MPH. It is no wonder that this is one of the best pitches in all of baseball, creating plenty of swords (ugly swings and misses for my older readers) with it.

Some other notable names in the top 25 are Aroldis Chapman, Tony Gonsolin (arguably the best splitter in the league), Jharel Cotton, and Hector Nerris. The changeup is unique compared to the rest of the pitches as it is not based off the movement of the pitch by itself. Instead, it relies on the difference of movement and speed from the fastball. Pitchers that are able to distinctly separate their fastball and changeup see the most success on the pitch, thus are scored the highest. Sinker dominant pitchers struggle to score well on the changeup due to the fact that their fastball is already sinking like a hard changeup. It is very difficult to separate the changeup even more from that, thus resulting in a lower score. This is where it is important to look at type of pitcher before making any decisions based on the changeup score.

Player vs Player Comparisons

Before we briefly talk about some brief points for the model, I want to go through some examples on how players can be evaluated against each other based on these metrics. Remember, Stuff + isn't the only thing that should be used to say one pitchers pitch is better than the other. It's a way for analysts, coaches, players, and fans to quantify it with a measurable statistic that can be used like any other. With all that said, let's get started on the comparisons.

Justin Verlander vs. Shane McClanahan

The two leaders in the AL Cy Young race as of 6/29. Both starters are pitching for playoff hopeful teams and are dominating. Verlander is enjoying a bounce back season after missing the entire 2021 season. While some might call it a hot start with expected stats being significantly higher (in a bad way) than his true numbers are showing, Verlander has stamped his name as one of the front runners of the AL Cy Young. McClanahan on the other hand struggled early in the season but has been dominant ever since, trying to will the Rays into a playoff spot with the tough AL East. In only his second season, the lefty is showing he can be a dominant starter for years to come. Let's take a look at the numbers for both these guys.

Both pitchers have dominant fastballs, but the model gives the slight edge to McClanahan. The entire off-speed arsenal is given to McClanahan as well as he has 4 pitches that are all significantly above average whereas Verlander has 2 pitches that are pretty much average and 1 that is below average. These numbers would back some notion that Verlander has gotten "lucky" to start the season. It will be interesting to see how these numbers are adjusted as the season goes along and to see where these two front runners line up towards the end of the season.

Patrick Corbin vs. Madison Bumgarner

Let's shift to two pitchers who had dominant histories that have fallen off that peak mountain. Patrick Corbin had an incredible season with the Diamondbacks in 2018 and he signed with the Nationals in 2019 where he helped win a World Series that year. However, since 2019 he has been hit extremely hard and signals the notion that it is incredibly difficult to succeed as a two-pitch starter. Madison Bumgarner enjoyed plenty of success with the Giants, being selected to 4 All-Star games and finished in the top 6 in Cy Young voting for 3 seasons. Bumgarner's success has not transferred to the Diamondbacks and looking at his stuff, it is easy to see why.

Patrick Corbin, a traditional fastball-slider pitcher, has begun to throw some other pitches at a small margin, but none of them providing any success. His best pitch in terms of score is the changeup, yet it still sits at 26% worse than league average. Bumgarner's best pitch in terms of score is the cutter which sits at 20% worse than league average. Just by looking at these numbers, it is easy to see why neither pitcher can find much success this year.

Taylor Rogers vs. Josh Hader

We'll pivot back to greener pastures with the two leaders in NL saves. Taylor Rogers has enjoyed success with the Padres after going to his first all star game with the Twins in 2021. He has already doubled his save count from last year and is looking at another all star appearance if he continues the success. Josh Hader has been the premier closer for the Brewers for years now and it is no shocker to see why.

Rogers wins the battle on stuff with a sinker that's 34% better than average and a league average slider. Hader has a changeup that is 30% better than league average with a league average fastball. Josh

Hader suffers from throwing a sinker which really moves like a 4-Seam Fastball (18 inches of carry and 8 inches of run). If his sinkers were correctly tagged as fastballs based on movement, I would imagine he would have an elite fastball, but because they are tagged as sinkers, it results in a below league average pitch. This is where pitches may not show well on the model but will still have success when used.

The Outlier

The player we are going to take a look at is the key reason why you need to be careful when using this metric exclusively to evaluate players. Tyler Rogers is a reliver for the San Francisco Giants that throws at a submarine slot. Because of this, his numbers don't show up very well.

His fastball is 95% worse than league average because it is slow and has the exact opposite movement. Despite the speed, his sinker and slider are only 36% and 45% worse than league average respectfully. Tyler Rogers is an effective relief pitcher because of this arm slot and movement that the model cannot account for. This is a rare instance where you have to judge the pitch based on other factors, because this cannot accurately reflect the quality of the pitch. In all cases, you shouldn't exclusively look at the Stuff + number for judgement, but in this case, you really shouldn't pay much attention to it at all.

Concluding Thoughts

Because we are still early in the season, the positions have plenty of room to change. The sample size of data, while immensely large to begin with is still only about a third of what it will be at the end of the season. More data will allow us to make even better, more accurate conclusions. However for the time being, over 300,000 pitches of data is plenty to accurately reflect pitchers performance with the model. Speaking of accuracy, this model has been validated among predictive and descriptive stats such as FIP, SIERA, WHIP, wOBA, etc. Furthermore, this model has been validated among other Stuff + models, Driveline's for example, and similar conclusions and scores about players have been found. This model is a little more conservative compared to other models, which is not the end of the world. This model still holds enough weight to be used in player evaluative settings, for player development, and for baseball fans of any team. This metric can even be used at the collegiate level as many schools are beginning to utilize pitch tracking systems such as Trackman and Yakkertech. This would be irrelevant at the high school level as players don't throw hard enough and they don't have the pitch tracking systems that would make it possible.