Get Nasty: Quantifying a Pitcher’s “Stuff”

This article was co-authord by Daanish Mulla (@DanMMulla)

A New York Times article by John Branch in October 2015 discussed the elusive definition of the pitching term “stuff”. Talk of “plus stuff” and feelings of “all the stuff being there” was scattered throughout the article. Despite interesting commentary discussing the ability for pitchers to over-power hitters, there was no true definition of the nastiness of a pitcher’s stuff.

Earlier this November, Eno Sarris wrote an article examining who had the best changeup in the 2015 season. This was evaluated by looking at the difference in speed and movement with respect to the pitcher’s fastball. This made us think, to truly quantify “stuff”, you would first need to understand what goes into a pitcher having a truly dominant repertoire.

Our definition of a pitcher’s “stuff”, or their overall nastiness, was based on three different factors: 1) fastball velocity; 2) change of velocity of a secondary pitch with respect to the fastball; and 3) movement with respect to the fastball. We downloaded all of FanGraphs’ PITCHf/x data from 2008 to 2015 to attempt solving this problem.

For a pitch to qualify for this analysis, it had to be thrown by an individual pitcher at a frequency equal to, or greater than, the average frequency for that pitch to be thrown throughout the entire data set. For example, in our data set, the curveball was thrown at an average of 12% of the time by all pitchers. Thus, a pitcher’s curveball was only considered if it was thrown at a frequency of greater than or equal to 12%. We then determined the maximum and minimum velocity for all eligible pitches for each pitcher. Working off of the fastball, we then determined the maximum change in movement in both the X direction, and the Z direction, for any qualifying pitches. We then calculated the maximum resultant movement for these values. Z-scores were then calculated and summed from the following factors to get a final pitcher “stuff” score: 1) maximum velocity; 2) change in velocity between maximum and minimum velocity; and 3) maximum resultant movement.

Here is an example as to how a pitcher with elite stuff performed in this analysis. David Price had a great year with the Blue Jays and Tigers. From FanGraphs data, his maximum pitch velocity was 94.1 mph, and the minimum pitch velocity was 85.2 mph – a difference of 8.9 mph. Working off the fastball, the greatest x direction break on a pitch was 15.1”, and the greatest z direction break was 10.9”.  This produced a resultant change in movement of 18.6”.

These values translated to a z scores for velocity, change in velocity, and resultant movement of 0.969, -0.08, 0.91, resulting in a stuff value of 1.80. Comparatively, another Blue Jays starter who struggled in 2015 was Drew Hutchinson. Hutchison had a fastball velocity of 92.4 mph, an offspeed pitch of 84.3 mph, an x direction break of 7.1, and a z direction break of 9.8. Corresponding z scores for velocity, change in velocity, and resultant break were 0.392, -0.24, -0.08, resulting in a stuff value of 0.1.

To break down how well our stuff rating was performing, we correlated stuff with K/9. Pitchers included in this analysis were all starting pitchers who pitched 90 innings in a season, between the 2008 and 2015 season. Average stuff and average K/9 was calculated during this time. Overall, the correlation was r = 0.42 (Figure 1). For the sake of these graphs, knuckleballers Tim Wakefield and R.A. Dickey were not included, as the stuff metric had them rated lower than -4 per season.

View post on

Figure 1. Stuff vs K/9, between the 2008 and 2015 MLB season.

Here’s the top 25 starting pitchers from the 2015 season ranked by their stuff. While overall, we think this is a good starting point for evaluating a pitcher’s repertoire, there are a few notable pitchers that the stuff calculation doesn’t seem to do justice. Chris Archer, who has had his slider called one of the best pitches in all of baseball, has only a 1.12 stuff value, and is ranked as having the 67th best stuff. Max Scherzer, who threw two no-hitters, is ranked as only having the 60th best stuff.

View post on

Table 1. Top 25 stuff for pitchers, with raw data on velocity and break

What’s worth stressing however, is that this metric serves to evaluate the individual pitches within their repertoire. There are pitchers which would be scouted to have the ability to throw hard, with lots of break. Pitching is clearly an art form that involves more than those two things, thus players like Mark Buerhle (-2.7), are clearly someone who has mastered the art of pitching, without having great stuff.  When comparing stuff against xFIP, correlation coefficients are smaller (r = -0.33) (Figure 2). Much like K/9 does not directly predict pitcher success, neither does stuff.

View post on

Figure 2. Stuff vs. xFIP, between the 2008 and 2015 season.

We believe there’s great use for this metric. We think this metric can provide insight into how stuff changes with age, how stuff changes after a pitcher is injured, and how it can let a coach know when a player has returned to pre-injury form, and how a pitcher’s consistency with their stuff relates to success. As with any ranking that appears on the FanGraphs website, we’re sure that there will be debate – however, we are looking forward to the input from the community into how we can improve this technique.


Branch, J. (2015). The Mysteries of Pitching, and All That ‘Stuff’. Posted online, October 3, 2015.

Sarris, E. (2015). The Best Changeups of the Year by Shape and Speed. Posted online, November 9, 2015.

Ergonomist (CCPE) and Injury Prevention researcher. I like science and baseball - the order depends on the day. Twitter: @DrMikeSonne

Newest Most Voted
Inline Feedbacks
View all comments
8 years ago

I really like where this is going. Nice work.

My suggestions:

If there is any way to quantify “lateness” of movement in this metric that could help separate the real bat missing offerings from pitches with more of a sweeping and less deceptive flight path.

Also please correct me if I am misinterpreting the method but Fastball movement is not helping one’s score here since the fastball is being used as the point of reference for the relative movement of other pitches: correct? If so some way to give credit for a rising/running/tailing/sinking fastball vs a “straighter” offering may also help improve the metric.

I look forward to reading more about this.

8 years ago
Reply to  Mike Sonne

Thanks for the in-depth reply. That makes sense about the FB/overall zone coverage. I wonder how a batter feels about a pitcher that throws a cutter and slider in the same “tunnel” vs a pitcher that throws a rising 4 seam and a hammer curve that break opposite ways. Do the pitches breaking opposite ways wreak more havoc or do two pitches that move in similar directions but differentiate themselves later in the total flight path cause more disruption of timing?

I love work that tries to answer these kinds of a questions; keep up the great work!

8 years ago

What about correlating ‘stuff’ with ERA-FIP or soft contact? Or even just RA9? When you get to measure real world inputs instead of outcomes, it’s cool to see if they are more predictive than the outcomes themselves.