Is Hitting Barrels a Skill?
And a mea culpa for Brandon Belt
Back on June 14, I wrote about how Brandon Belt was having so much success during the 2018 season, despite not possessing a high average exit velocity, by hitting for a high average launch angle. I argued that, in doing so, Belt was optimizing his batted-ball profile.
In my article, I noted that, at the time, Belt had a .401 wOBA and a .492 batted-ball wOBA, which ranked 10th and 11th, respectively, among hitters with at least 100 plate appearances (PA). For comparison’s sake, I also noted that, at the time, Belt’s 87.6-mph average exit velocity ranked 233rd among the same sample of hitters, while his 34.8% hard-hit rate ranked 156th among hitters with at least 100 batted-ball events (BBE).
Finally, I concluded that, despite the relative dearth of hard-hit balls, Belt’s 17.7% barrels-per-BBE rate, which ranked sixth among hitters with at least 100 BBE at the time, was a product of his high average launch angle (22.9 degrees, highest among hitters with at least 100 PA at the time). Moreover, I lauded Belt’s ability to selectively hit his hard-hit balls for barrels, which I demonstrated by inventing a stat (barrels-per-hard-hit rate), at which Belt’s 51.0% was the highest among hitters with at least 100 BBE at the time.
That being said, here are Belt’s slash-line stats (plus wOBA) on June 14 and at the end of the season:
So, what happened to cause such a drastic decrease in productivity?
Before diving into the stats, it should be mentioned that, between March 29 (the start of the season) and June 1, Belt played in every Giants game. Starting on June 2, Belt was out for two weeks because of appendicitis. Then, starting on July 26, Belt was sidelined for 17 games, this time because of a hyper-extended right knee. Finally, Belt had season-ending surgery on that same knee in mid-September, playing his final game of the season on September 14.
So, in large part, the answer is injuries. But, is there something more that we can glean from the stats?
Despite his strikeout-rate remaining exactly the same, Belt’s walk-rate decreased after his first stint on the DL. That’s obviously part of the answer of why Belt’s productivity decreased, but we’re not here for that. We’re here for what happened when the ball was put in play.
Here’s where the main culprit lies. Because of Belt’s battles with injuries (presumably), his average exit velocity decreased from 87.6 mph before June 2 to 84.8 mph after June 2, with his average launch angle remaining relatively the same. This 2.8-mph decrease in average exit velocity is seemingly the cause for the 206-pt. decrease in wOBA on batted balls and the 173-pt. decrease in expected-wOBA on batted balls.
However, because I don’t trust average exit velocity or average launch angle, I dug a little bit deeper:
As expected with a decrease in average exit velocity, Belt’s hard-hit rate had a corresponding decrease. However, even though Belt’s average launch angle remained relatively the same, his groundball rate increased, with a corresponding decrease in airball rate. The increase in groundball rate probably explains the increase in Belt’s batted-ball “unluckiness”, as measured by difference between xwOBA on BB and wOBA on BB, from 19 pts. to 52 pts. via hitting into the shift more often.
Given Belt’s 6.0-pt. (17.2%) decrease in hard-hit rate and 4.6-pt. (5.9%) decrease in airball rate, I would expect Belt’s barrel rate to have decreased from 17.7% to ~13.8%. Yet, Belt’s barrel rate decreased to 3.9% after June 2, in large part because Belt’s barrel-per-hard-hit rate decreased from 51.0% to 13.6%.
Was Belt’s greater-than-expected decrease in Brl/HH% (and subsequent greater-than-expected decrease in Brls/BBE%) simply a regression to the mean or did Belt’s barrel-hitting skill actually worsen because of the injuries?
There are two ways that I interpret Belt’s decrease in Brl/HH%. The first is that it is simply a regression to the mean, for 51.0% was unsustainable over the course of a full season. The second is that, as average exit velocity and hard-hit rate decrease, the first batted-balls to go are barrels, so barrel rate and Brl/HH% would subsequently decrease at a disproportionately larger rate.
There is evidence that the second interpretation is more correct, for there are fairly strong correlations between Brl/HH% and average exit velocity (0.60) and hard-hit rate (0.65) since 2015 among hitters with at least 150 BBE in consecutive seasons. Although, there may be selection bias involved, for there is a positive correlation (0.28) between average exit velocity and average launch angle, suggesting that hitters who hit the ball harder, on average, are more likely to hit the ball at higher launch angles at all exit velocities.
Furthermore, here are the 2015–18 year-to-year correlations for the stats in question:
I would like to this time to say that if this is at all confusing to you, then you’re not alone. I’m confusing myself a little bit as I’m writing this.
But I think the main takeaway should be that, as suggested by the above correlations, how hard and how high a hitter hits the ball on average are skills. Hitting barrels is also a skill, as suggested by the strong year-to-year correlations in both Brls/BBE% and Brl/HH%, in part because exit velocity and launch angle are skills, but also in part because hitting the ball hard at a higher launch angle is also a skill.
Furthermore, since 2015, the correlation between Brls/BBE% and Brl/HH% is 0.96, suggesting that barrel rate will fluctuate among hitters even when hard-hit rate is held constant. Also, since 2015, average launch angle and hard-hit rate do not explain all of the variation in Brls/BBE% (adjusted r-squared = 0.70), while average launch angle does not explain all of the variation in Brl/HH% (adusted r-squared = 0.30).
Put another way, hitting barrels is not just a function of average exit velocity and average launch angle. Rather, to some extent, it is a function of the intersection of the two.
For instance, Danny Valencia and Nick Markakis finished the 2018 season with similar average exit velocities (90.7-mph and 90.6-mph, respectively) and similar hard-hit rates (41.1% and 41.3%, respectively). Yet, even though Markakis (10.2 deg.) had a higher average launch angle than did Valencia (9.0 deg.), because Valencia (94.5 mph) hit the ball harder in the air, on average, than did Markakis (92.3 mph), Valencia had the higher Brl/HH% (19% vs. 8%) and, thus, the higher Brls/BBE% (7.7% vs. 3.2%).
That being said, here are the hitters who exhibited the most “barrel ability” in 2018, as measured by the difference between actual Brls/BBE% and expected Brls/BBE% based upon launch angle and hard-hit rate…
… and least “barrel ability”:
Nevertheless, here are the hitters who were most “barrel lucky” in 2018, as measured by the difference between actual Brl/HH% and expected Brl/HH% based upon launch angle, FB/LD exit velocity, and GB exit velocity…
… and least “barrel lucky”:
Thank you for reading, and tune in next time when my topic will be… something different
