Fastball Velocity, Fastball Usage, and All That Fun Stuff

For the better part of this decade, we’ve repeatedly published an article you can more or less predict. Nearly every year, a version of the same idea gets published. “You’re never going to believe it,” the article starts, “but fastballs got faster again this year.” There are usually some GIFs, maybe a winking joke about how we write this article every year and it keeps being true, and bam, 1,500 words out the door. Oh yeah! There’s also a kicker: “Fastballs keep getting thrown less frequently, too.”

Normally, I’d be writing that article again this year. There’s just one problem: four-seam fastballs didn’t get faster this year; in fact, they’ve been plateauing for a few years. This year’s four-seamers checked in at an average velocity of 93.9 mph. Adjusting for time of year (I used only data from August onward in each season so that we didn’t have any weather effects unique to 2020), here are the last five years of four-seam velocity:

The 2019 season was the fastest on record, and 2020 fell short of that mark. In fact, the last five years look overall unchanged. Look instead at sinkers, though, and you’ll see some velocity improvement:

Which one should we believe? Four-seamers are more common than sinkers, so the blended average looks like this:

Okay, so fastballs didn’t get any faster this year. Sinkers did, and that’s interesting for sure, but at the highest level, it feels like the inexorable march towards higher velocity might have stalled for the moment.

The other yearly trend, though, shows no signs of slowing. Take a look at the share of fastballs as a percentage of all pitches. I’ll even zoom this one out to 2008 so that you can get a sense of scale:

For nine years, pitchers threw a near-constant share of fastballs. From 2017 onward, that number has cratered, and 2020 represented a new low. There’s no reason to expect this trend to stop in the way that velocity did, either. Easy velocity gains have some limit: At some point, a lot of pitchers will have gone to pitching labs to improve their mechanics and conditioning, and teams will have tapped all the hardest throwers from their farm systems. On the other hand, all you have to do to throw fewer fastballs is throw fewer fastballs.

I’m resisting the urge to make some broad sweeping statement about the future course of the game, but it certainly doesn’t feel like we’ve reached an equilibrium rate of fastballs. Batters still want to see a fastball every pitch. Per our pitch values, hitters were 234 runs above average against fastballs and 234 runs below average against all other pitches.

That differential hasn’t moved much, even as pitchers throw fewer fastballs — in 2015, it stood at 594 runs against fastballs and -594 against all other pitches, which is roughly the same after pro-rating 2020 to a full season. In other words, batters keep getting better at hitting fastballs, which makes them less and less attractive for pitchers to throw, which keeps the overall value lost via fastballs the same even as it increases per pitch thrown.

Where is a fair balance? It’s hard to say, because we’re extrapolating outside the range of historical pitch mixes. Additionally, it probably doesn’t make sense to throw fewer and fewer fastballs until those pitch values both get to zero. There are ancillary benefits to fastballs that aren’t covered in pitch values. They set up other pitches; changeups wouldn’t make too much sense if they were thrown more frequently than fastballs, for example. They also limit workloads on pitchers, as fastballs are far more likely to be put into play than secondary pitches, and you can’t exactly go eight sliders deep on every batter you face. There are other reasons, too; suffice it to say that it’s possible for a fastball to be a good pitch without a positive run value.

If I don’t have any predictions on that front, then, what’s the rest of this article going to be? I had a theory about the velocity drop that might account for the decline in four-seam velocity in 2020 without truly predicting a plateau. There are two ways that average fastball can increase. First, the league can find new pitchers who throw hard. Jamie Moyer isn’t in the big leagues anymore, but Brusdar Graterol is. Every fastball thrown by a young flamethrower instead of an old veteran is a push higher in average velocity.

Second, pitchers don’t throw the same speed forever. Zack Greinke used to touch the upper 90s, and now lives in the 80s. Aging comes for everyone, and it tends to push velocity lower. In theory, both of these factors can flow in either direction: There’s no rule that the pitchers debuting need to throw harder than the pitchers who stop appearing, and there’s also no rule that the same pitcher can’t gain velocity. In practice, though, the replacement effect pushes average velocity higher while the aging effect pushes it lower.

This season threw a wrench into the works. Pitchers came into the 2020 campaign after an extended layoff and abbreviated ramp-up period. We saw, in injuries and workload, how that affected them. It also affected their average velocity, naturally enough, and the data shows it clearly.

I took the same dataset I used to find league-wide velocity: four-seam fastballs thrown in August and September. Then, I broke it down by pitcher and looked at year-to-year changes. For example, Oliver Pérez averaged 89.7 mph on his fastball in 2020. In 2019, that stood at 91.5, which means he lost 1.8 ticks (accounting only for August and September games). In 2018, he was at 91 mph on average, which means he gained half a tick from ’18 to ’19.

I ran this calculation for every pitcher in baseball. The database I used goes back to the start of the Statcast era (2015), which means that I only have changes starting from 2016 on. The simple average change in fastball velocity for returning pitchers took a big downtick this year:

Again, not a surprise: pitchers were out of training in 2020, and as a result they didn’t throw as hard. It’s not exactly rocket science. If you weight it by the number of pitches thrown in 2020, the effect persists, though it’s not quite as large:

What does this mean for the future of articles about average fastball speed? We’re going to have to wait for 2021 to find out. If returning pitchers had shown exactly unchanged fastball velocity in 2020, four-seam velocity wouldn’t have changed at all league-wide. We’d still be at a peak, rather than just below one. That’s not to say that 2020 was necessarily a blip, or that fastball velocity will go back up in 2021 — after all, 2019 was the first year with gains from one year to the next — but it does suggest caution in calling a peak.

Will a more predictable schedule see a return to 2019’s velocity gains? Was 2019 the outlier rather than 2020? It’s too soon to know, and in fact, we might need a month or two of 2021 data to find out for sure. The two paired harbingers of modern pitchers, fewer fastballs and faster ones, might be coming unpaired. We’ll find out soon enough.