When reading another Fangraphs piece a while back highlighting Cole Ragans, I was intrigued by a note from Ben Clemens's analysis of Ragans's Changeup and the belief that it, despite no way to quantify it, was helping the rest of the arsenal. This led me to wondering how often pitches perform better relative to expectation as a result of effects from one "signature" pitch in an arsenal?
Comparing Fangraphs' pitch grades (expectation) to actual performance, for this we’ll use Statcast Run Value per 100 pitches (RV/100), to identify pitchers with arsenals containing pitches with below-average grades and above-average performance, and one “signature” pitch. Instead of choosing one specific set of criteria to define “signature”, we’re looking at different ways to create this definition, and evaluating how the results change.
A couple of notes for data collection/prep:
Looking at individual pitcher seasons from 2021-2023
Pitchers included within a given season must meet a 20 innings pitched minimum qualifier
Each individual pitch type must have at least 100 pitches thrown in a given season
Pitchers who threw at least 100 Sliders and 100 Sweepers within a single season were REMOVED from the sample due to not having proper distinction in Fangraphs' pitch grades
i.e. Yu Darvish
Solely looking at arsenal comprised of:
At least 3 pitches TOTAL
At least 1 below-average pitch grade
One “signature” pitch
Below are the proportions of pitchers in each population that saw ALL of their below-average pitches in their arsenal perform with a positive RV/100.
When defining “signature” as a grade > 100 AND RV/100 > 0
Stuff+ - 23.8% (336 pitchers)
Pitching+ - 23.4% (346 pitchers)
“Signature” as grade > 1 standard deviation above pitch type average AND RV/100 > 0
Stuff+ - 31.1% (135 pitchers)
Pitching+ - 32.5% (154 pitchers)
“Signature” as grade > 2 standard deviations above pitch type average AND RV/100 > 0 (small sample warning)
Stuff+ - 38.9% (36 pitchers)
Pitching+ - 30.4% (23 pitchers)
What are we taking away from this? There’s a solid amount of pitchers that outperform bad pitch grades when also throwing ONE really good pitch! It’s even more pronounced when the criteria for defining that “signature” pitch is raised to a higher standard, something not necessarily surprising, but nonetheless interesting to see the percentages play out that way.
With that being said, a lot of these successful pitchers have just ONE bad pitch in the arsenal, and obviously we can’t conclude any sort of direct effect and influence of the signature pitches on the bad ones, so we’re far from drawing any meaningful conclusions about the influence of signature pitches on an entire arsenal.
But I think this opens up the doors for some more interesting questions. Specifically, can we look into the success cases to better understand WHY they are able to outperform bad pitch grades, and what might some similarities be between these pitchers, and differences from those that throw signature pitches but still have poor performance on pitches with bad grades. Is there a pattern or strategy to the way the successful pitchers are using each pitch in the arsenal from an overall frequency or sequencing standpoint?
Why might it be important to explore questions like the ones in this post, and further ones that I just mentioned? I think the obsession of pitch grades in recent years has placed a priority on maximizing stuff quality across an entire arsenal. But what if instead it’s more important, or more efficient, to focus on maximizing the grades of one specific pitch in the arsenal, and the deployment strategy between a signature pitch and the rest of the arsenal? I’m not necessarily saying I believe one side more than the other, and not that there are necessarily “sides” to pick between to begin with, but rather more interested in elaborating on what we think we currently know about pitch design.
Another shoutout to Ben Clemens for, perhaps accidentally, triggering another interesting question.