Image credit: © Geoff Burke-Imagn Images
A few weeks ago, I discussed the option of moving beyond current metrics for summarizing exit velocity, such as the raw mean and the 90th percentile. The reasons for this include that (1) the exit velocity distribution is heavily left-skewed, so the raw mean is missing important information; (2) the 90th percentile tells you just that: the 90th percentile, not the rest of the distribution; and (3) neither metric allows you to best summarize and project a hitter’s entire velocity distribution, which is important if you want to understand how effectively they deploy exit velocity over the course of a (projected) season, for fantasy or other roster-building purposes.
Exit velocity generally involves the expenditure of maximum average athletic effort, and I explained in the previous article why the skew normal distribution allows us to help address these issues, naturally accommodating a heavy skew without losing fitting power. The skew normal thus allows us, at least in theory, to summarize all 100 percentiles, not just one. And at any time, we can issue a proposed distribution of expected exit velocities for any player, making it easier to interact with launch angle and other batted-ball characteristics to better understand what a player is doing to get the results that they are.
What that article did not do, however, was discuss specific examples of when the distinctions between these metrics might matter. Today, we focus on that question. Going forward, I will refer to a player’s skew mean exit velocity as their “Deserved Exit Velocity” or dEV, as the other term is a bit of a mouthful, and would be confusing given forthcoming adjustments to other batted ball measurements that do not vary in a skewed manner.
To illustrate the use case, let’s compare three exit velocity densities from the 2024 season: (1) the league-wide average distribution; (2) the distribution for Pete Alonso; and (3) the distribution for Luis Arraez:
No one would argue these distributions are the same. Yet, existing metrics presume the opposite, that player exit velocity distributions differ only by the extent to which they are shifted right or left along the exit velocity (launch speed) spectrum. As such, these metrics ignore how certain players might goose their exit velocity on the high end while having too much below average contact, or how players with lower high-end velocities might backstop that deficiency through greater concentration around their true average. These differences shouldn’t be ignored, and now we no longer have to.
Let’s compare dEV vs. other values measurements for these and other MLB players.
Batters
As usual, our models provide uncertainties around our estimates, but instead of providing more numbers, we’ll use the uncertainties to select only those batters whose dEV was more than three standard deviations away from the average value. In other words, we have very high confidence in these dEV values. We will provide both the underlying values for 90th-percentile exit velocity (EV90) and then compare percentiles:
Table 1: dEV Minus EV90 Laggards, by Percentile
The batters at the top of Table 1 tend to be known both for hitting the ball hard and for not having the corresponding results you would expect. Some of this has to do with launch angle issues, but Pete Alonso (50th percentile dEV versus 88th percentile EV90) has struggled to generate free-agent traction this offseason, and Jo Adell can never seem to get over the hump. Jorge Soler’s career has been a land of contrasts, and Christian Walker (63rd percentile dEV versus 83rd percentile EV90) is a name Astros fans cannot be happy to see here. In any event, by dEV, these players may be worth a second look if that 90th percentile exit velocity might be viewing them more positively than it should.
Table 2: dEV Minus EV90 Leaders, by Percentile
The batters in Table 2, on the other hand, are those dEV sees as being sold short. Brandon Nimmo stands above the rest, with an 85th percentile dEV versus just an average EV90. Jordan Walker and MJ Melendez also get the nod for the way they generate exit velocity although, again…launch angle issues. Finally, dEV picks up Luis Arraez and Stephen Kwan, acknowledging that while their high-end exit velocity may be absent, their concentration in other areas is robust, maintaining an average-ish exit velocity when other batters start to drop off fast.
Pitchers
Pitchers are where dEV particularly stands out, providing notably better reliability year-to-year than average launch angle, and significantly better reliability than EV90, which appears to be a poor choice for pitchers. This probably is because pitchers do not affect the skew of the exit velocity distribution, just the mean and, to some extent, the standard deviation.
Pitchers also affect exit velocity less than batters in general (batters had just over 2.5 times the influence of pitchers in 2024, we estimate), so we will lower our uncertainty filter a bit and require values to be two standard deviations away from average rather than three. (This is still a high degree of certainty.) We will also maintain the notion of a higher percentile being better, so in this case, a high percentile means that a pitcher has a lower exit velocity location, whereas with batters it was the opposite. Finally we will use the mean rather than the 90th-percentile exit velocity, because the former better evaluates pitcher exit velocity control.
Table 3: dEV Minus ev_mean Laggards, by Percentile
Ryan Yarbrough jumps off the page; his raw mean exit velocity is one of the best in the game, but his dEV is merely middle of the tier, a difference consistent with his varying performance over the years. Raisel Iglesias encounters the same skepticism, with dEV being far less impressed with his overall distribution than his raw midpoint. It fortunately didn’t matter for Iglesias, though, who at 34 continued to minimize free passes, finding substantial, if slowly diminishing, success. José Soriano is flagged also, although his exceptional groundball rate somewhat limits the damage. The Rogers brothers interestingly both show up on the list as well.
Table 4: dEV Minus ev_mean Leaders, by Percentile
The differences are less stark on our Underrated list, although some notable names nonetheless make the list. Of course, Kyle Hendricks is near the top, despite his other challenges, showing that exit-velocity management alone is not sufficient for success. The remarkable Max Fried, fresh off signing a $218 million deal with the Yankees, gets a big vote of confidence as well; he’s skilled at suppressing contact in multiple ways at once. The appearances of Zach Wheeler and Justin Steele also feel notable, given their sustained success.
As always, remember that just because two metrics disagree does not mean that one is “wrong,” just that they may be asking slightly different questions. Metrics cued to raw exit velocity (the mean or 90th percentile) presume that every player’s exit velocity distribution is essentially the same; dEV does not. However, since dEV for pitchers is notably more reliable than raw mean exit velocity, and much more reliable than 90th-percentile exit velocity, dEV may be a better choice to evaluate pitcher exit velocity when it is available.
Next Steps
We will continue to roll out adjusted batted-ball metrics in the hope of shedding more light on how interesting players might be achieving their results. In the meantime, we welcome your feedback, including whether you would like to see dEV as an available column on our leaderboards this season.
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.
Subscribe now
