Should You Believe Exit Velocity Breakouts?

Rick Osentoski-USA TODAY Sports

For the past few weeks, I’ve been delving into exit velocity readings in an attempt to find out what really matters and what’s just noise. I found that 95th-percentile exit velocity and contact rate are the two stickiest metrics from one year to the next, with exit velocity slightly more likely to remain the same from one year to the next.

Of course, that doesn’t mean it can’t change. In fact, players change their top-end power readings by a good amount every year. Sure, any individual player might be unlikely to do it, but there are tons of players in baseball. I found that only 4% of hitters change their 95th-percentile exit velocity (EV95) by one standard deviation from one year to the next, but 408 batters put at least 100 batted balls into play in 2022. Four percent of 408 is a lot more than zero.

With that in mind, I thought I’d take an inventory of those exit velocity changers and see what their improvement meant going forward. To do so, I created two groups: hitters whose EV95 improved by at least half a standard deviation from one year to the next, and the opposite, hitters whose EV95 declined by at least half a standard deviation. I picked half an SD instead of an entire one to bulk up the sample size.

That left me with 190 EV95 improvers and 251 decliners. From there, I added another filter: I only kept players who also played in the subsequent season. That left me with 139 improvers and 187 decliners. I then looked at those groups to see what happened the year after their big moves.

Let’s start with the EV95 improvers, because improving is more interesting than declining, at least in the pursuit of hitters who might already be showing raw skills that haven’t yet translated to production. On average, they improved their EV95 by 2.2 mph in year one. I did a little regression work in the background – that amount of EV95 improvement works out to a 13 point expected improvement in wOBA holding all other metrics constant.

In the subsequent year, our EV95 improvers lost 0.9 mph. In other words, they gave back a little bit less than half of their improvement. That’s not too shabby; you’d expect a fair amount of regression to the mean just mathematically, due to the sample I selected. It’s reasonable to assume that hitters who recorded a higher measurement in year two than in year one were, on average, a little bit lucky.

It’s like how nearly every professional poker player has run above expected value in their career – the ones who had worse-than-average luck at the beginning of their careers are less likely to stick around. If you take an average of a successful group, that group will almost always have above-average luck. If two hitters each improved their “true talent” EV95 by a number just short of qualifying for our study, but one had an extra 0.5 mph boost due to luck and one had a 0.5 mph headwind due to bad luck, my criteria will only capture the lucky one, and that hitter will have a natural decline in the subsequent year.

On the other side of our study, decliners didn’t rebound quite so much. The average decline among hitters whose EV95 fell by half a standard deviation or more was 2.2 mph, the exact opposite of our improvers, which is convenient. But they only rebounded by 0.6 mph, a slightly smaller magnitude. There are two possible explanations for that: random chance, or an overall headwind across the population.

I’m actually partial to the second answer. The average change in EV95 from one year to the next among all qualified hitters is slightly negative, a decline of roughly 0.2 mph per year. My ex-post explanation for that is that EV95 is pretty close to measuring rotational energy; how much force you impart on the ball when you hit it squarely, more or less. That’s a young man’s game, and every player in our sample is exactly one year older in the second year than in the first (thanks, math!). The small decline is likely picking up an aging effect, though I haven’t looked into it with enough nuance to say for certain.

Another way of looking at this is to ask what percentage of hitters held their gains (or losses) from one year to the next. In other words, if you see a hitter make a leap in their raw power measurables, what are the odds that they hold those gains in their next season? Well, 26.6% of EV95 improvers improved again the next year, and 35.8% of EV95 decliners saw no exit velocity bounce back in the subsequent season.

What does this mean for you, the reader at home who is likely reading these articles with an eye towards fantasy baseball? To me, it’s simply a useful sanity check. If you’re interested in a hitter exclusively because they showed a power pop last year, you should mentally discount the gain somewhat. Likewise, if you’re trying to do some bargain hunting and acquire hitters who had rotten years, don’t count on them to recover all the way to where they were before, even if you do give them credit for a small increase.

Another way of thinking about those power decliners: only 10% of hitters whose EV95’s declined by at least 0.5 standard deviations in one year saw a full rebound in the next year. In other words, big declines are mostly real, or at least partially real. I’d be far more pessimistic about hitters who saw both production and raw number declines than ones who merely saw a downtick in production. If I were going after one of those guys, I’d want to find one with a good reason for their decline – an injury, most likely.

To finish this out for today, here’s a list of hitters who saw the biggest increases and decreases in their EV95 marks from 2021 to 2022. I can’t tell you which of them will have great years, but I can tell you that all of them are interesting for one reason or another:

EV95 Gainers, 2021-22

EV95 Decliners, 2021-22

That’s all I’ve got for today, but not to worry: it’s still the middle of February, and we’re still at a point in the baseball calendar when it’s quite difficult to come up with interesting topics. So tune in later this week for another iteration of “what can exit velocity tell us,” because I’m going to mine this vein for all it’s worth.

Ben is a writer at FanGraphs. He can be found on Twitter @_Ben_Clemens.

Newest Most Voted
Inline Feedbacks
View all comments
1 year ago

This is interesting and pretty much what one would expect.

From the point of view of identifying “breakout” players I’m wondering whether it adds anything to simply taking “result” stats like ISO. I’m guessing restricting the sample to those with “breakout” EV95 but no ISO increase would result in a tiny sample

I also wonder whether the breakout EV95 but same ISO just means the player hits GB harder. Is it really helpful to identify players who hit GB harder?

Last edited 1 year ago by Ivan_Grushenko