Another Post About Fastball Velocity

If you felt like you hadn’t gotten enough fastball velocity analysis at FanGraphs in the past week, do I have a treat for you! Last week, Kevin Goldstein expounded on the importance of the shape of a fastball in determining its effectiveness, making it clear that velocity is the driving factor in evaluating a fastball, but that deviating from a “normal shape” (interpreted as an inch of ride for every inch of run) can lead to missing more barrels. Devan Fink, meanwhile, showed that fastball velocities have increased across the league to start the year, which is especially striking since fastball velocities are usually at their nadir in April. Devan also highlighted pitchers who have seen an especially large bump. Finally, on Friday, Jake Mailhot took a look at Chris Paddack’s first start with a focus on the righty’s fastball after a disappointing 2020 season.

I recommend reading all those pieces, if you haven’t already. Kevin and Jake give credence to the idea that fastball shape is an essential factor in a good fastball. But I want to focus on Devan’s article on fastball velocity and how it seems to be increasing again in 2021, as it has every season since 2008. One can surmise that this is a product of pitchers acknowledging the importance of velocity (thus training with gains in mind) and teams giving more innings to pitchers who, by and large, throw harder. Velocity obviously matters, but how much?

To find out, I investigated the effects of year-over-year changes in fastball velocity on both the run values of those fastballs and the overall performance (in terms of wOBA allowed). I also wanted to see if changes in fastball velocity had more of an effect on pitchers who leaned into their heaters more. I looked at pitches classified as either four-seamers or sinkers by Baseball Savant. The run values are from Bill Petti’s Statcast scraper function in his baseballR package. For the changes in velocity, run values, and wOBA allowed, I looked at every pair of pitcher seasons since 2015 where the pitcher threw at least 250 fastballs each season.

The following represents the correlations between changes in fastball velocity and changes in the run value added per 100 fastballs (from the perspective of the pitcher) and in overall pitcher wOBA allowed, with each point weighted by the number of fastballs thrown by the pitcher:

There isn’t much to glean from this. Overall performance is a little bit more influenced by changes in fastball velocity than the actual fastball run values. But I think I can attribute some of this to the lack of year-over-year stickiness for run values, which are context dependent, as opposed to wOBA. So there is only a tiny amount of signal coming from perturbations in fastball velocity.

What about pitchers who depend on their fastball more? I conducted the same analysis, but this time, I fit the same simple linear models to buckets of pitchers. The buckets were determined by the rate at which the pitchers threw their fastballs, in increments of 10% (of their total pitches thrown).

Given the lack of a sample in the more extreme buckets in either direction, I would not place much value in those results. But for pitchers who use their fastballs more than a cursory amount, there are some stronger correlations between velocity changes and the two quantities of interest. Nevertheless, the relationships are not as strong as I expected. There is a relationship in either case — more so for pitchers who throw at least 30% fastballs — but even there is still a lot of noise.

There are copious amounts of past research on the importance of fastball velocity in the context of understanding both the pitch’s value itself and its affect on overall results. Almost three years ago, Alex Chamberlain modeled swinging strike rates and groundball rates for all pitch types and found that velocity, in a vacuum, was not sufficient to determine the merits of a pitch. He argues fastball velocity is useful in conjunction with other pitch characteristics (pitch movement and the velocity of the given pitch) but alone, it cannot tell much about the quality of the pitch. Using a gradient-boosted decision tree model, Kevin Dorosh, Julia Prusaczyk and Sara Stokesbury, concluded back in 2016 that for fastballs, velocity was “of middle importance” when evaluating the quality of a pitch in terms of pitch values. Daniel Epstein at Beyond the Box Score found there was little relationship at the team and player level between hitter wRC+ and opponent fastball velocity. And Mike Fast came to the conclusion that marginal gains in fastball velocities better explain the changes in run suppression for higher velocities than lower velocities. His theory was that pitchers who pitch in a lower velocity band get to the majors by virtue of adding value through other aspects of pitching.

For this analysis I chose to bin pitchers by fastball usage, as opposed to velocity, but my inclination would be that fastball usage and velocity are tied together. This is not really the case, as Ben Clemens outlines in this piece. Relievers, who tend to throw harder than starters, are increasing their rates of non-fastball pitches just like starters. This would lead me to believe that while the threat of a high-velocity fastball in a pitcher’s arsenal will always be important, an increased reliance on breaking and offspeed pitches will make those pitch characteristics more integral in explaining run prevention and pitch value.

So while velocity certainly matters and is an easy aspect of pitching to understand and monitor, fastball shape has more nuances, and we don’t have a firm grasp on what makes the shape of a fastball good or bad. A good fastball shape can depend on a pitcher’s arsenal, his arm slot, his height, and how well he can impart spin on the ball, and that’s without considering how well a pitcher locates his fastball as a result of those contextual factors. In general, losing velocity is bad and gaining velocity is good (hard to believe, I know), and that information is a valuable starting point in pitch evaluation. But if we want the full picture on the merits of a pitcher throwing his fastball, we need much more than just a velocity reading.