Why Matt Carpenter’s Production Is Misleading (and Complicated)
There are two hitters I would like to introduce. The first, Player A, has been described in terms of the classic trio of statistics: average, on-base percentage, and slugging. The second, Player B, has been described in terms of modern metrics like Exit Velocity and Barrel rate. Take a look at their numbers and try to see who’s better:
Player A: .081/.205/.162
Player B: 95.4 mph Exit Velocity, 63.6% Hard-Hit rate, 27.3% Barrel rate
Not much of a competition, right? Without additional context, you probably chose Player B in a heartbeat. Player A’s appalling triple-slash makes him a DFA candidate. Player B, on the other hand, looks like a hitting genius! Those numbers and rates would place him well above the 95th percentile of all major leaguers. The twist, of course, is that these two hitters are in fact the same person: Matt Carpenter, veteran infielder for the St. Louis Cardinals.
Traditional and modern metrics do disagree at times, but the disparity between them is seldom this wide. Through 18 games, Carpenter’s efforts to clobber the ball have not translated into actual results, much to the chagrin of Cardinals fans. There’s having a stretch of bad luck, then there’s hitting below .100. Is there something else we’re missing?
Let’s investigate and find out. Already, there’s a glaring problem with Carpenter’s production. While his numbers on contact are excellent, he hasn’t been making regular contact. His K% and Whiff% rank in the fifth and 10th percentiles, respectively. Baseball Savant’s metrics, which are calculated on a per BBE (batted ball event) basis, don’t reflect those issues. As a demonstration, I determined Carpenter’s Barrel and Hard-Hit rates based on his per PA results instead. Doing so casts him in a less flattering light:
Metric | per BBE | per PA |
---|---|---|
Barrel% | 27.3% | 13.6% |
Hard-Hit% | 63.6% | 32.6% |
These new rates wouldn’t look so salient on a Baseball Savant page. For a hitter who’s whiffing often, a change in the denominator has an outsized effect on how he’s portrayed. Carpenter has 22 batted balls as opposed to 44 plate appearances, 16 of which ended in a strikeout. That doesn’t necessarily mean we should switch to per PA metrics – whiff and strikeout numbers are easily accessible – but it’s easy to ignore the negatives and focus on the positive, especially as fans.
But sure, the fact remains that Carpenter has hit six barrels. Of the 336 hitters to record at least one, only 37 have more than six. His achievement is nothing to sneeze at. The fact also remains that those six barrels have resulted in a single hit (a home run, at least) and five outs. Probability-wise, it seems bonkers. There’s another caveat, though, which provides insight into why it’s still bonkers, but not as bonkers as you might think. That begins by reviewing the definition of a “Barrel.” Here’s a succinct one straight from the source:
“To be Barreled, a batted ball requires an exit velocity of at least 98 mph. At that speed, balls struck with a launch angle between 26-30 degrees always garner Barreled classification. For every mph over 98, the range of launch angles expands.”
Reading this unlocks a key observation: Not all barrels are created equal. It’s a cliche within the sabermetric community, but it’s true nonetheless. A 98 mph, 30-degree batted ball receives the same classification as a 115 mph, 30-degree rocket. For example, this mammoth Shohei Ohtani blast is a barrel:
But so is this Matt Carpenter fly out:
You can see where I’m going here. When Carpenter has managed to make contact this season, even his best instances have been relatively lackluster. Here’s how they’ve looked so far, according to a few important metrics:
Date | EV (mph) | LA (deg) | xwOBA |
---|---|---|---|
4/21/21 | 101 | 26 | 1.274 |
4/16/21 | 99.8 | 33 | .965 |
4/13/21 | 106.9 | 40 | 1.189 |
4/11/21 | 98.5 | 26 | .993 |
4/8/21 | 106.6 | 38 | 1.432 |
4/6/21 | 101.4 | 30 | 1.310 |
Carpenter’s barrels are those kids who barely pass the height requirement for a roller coaster ride. None of them are above 110 mph. Two of them are above 36 degrees, at which point it’s difficult to be productive, but they’re classified as barrels because of their exit velocities. Still, by xwOBA, these are some of the best outcomes a hitter can muster. So why is Carpenter not reaping the benefits?
Let’s consider the other four barrels that were hit under 36 degrees. They’re situated in the optimal launch angle range, at least – wOBA by launch angle tends to peak around 25 degrees, then tails off to the left or right of that mark. But what isn’t mentioned as often is that the optimal launch angle for a hitter changes depending on his Max EV, a measure of raw power. Courtesy of Connor Kurcon, below is a comparison between hitters whose Max EVs are below 110 mph and hitters whose Max EVs are above that mark:
Good example of how an "ideal" launch angle distribution isn't the same for every type of hitter.
wOBA by LA for hitters with an in-season max EV of 110+ (good proxy for power) vs. hitters with an in-season max EV of <110 pic.twitter.com/7v0AJ0tMJw
— Connor Kurcon (@ckurcon) February 17, 2021
For hitters who possess middling raw power, the sweet spot that exists in the 20 to 30-degree range is significantly diminished. Their best bet is to thrive on line drives, which are equally beneficial for both groups. Carpenter isn’t just unlucky; he’s also hitting the types of fly ball barrels that are unsuited to his skills. That only makes it harder to achieve consistent success.
His misfortune doesn’t end here, though. Earlier this month, Ben Clemens and I each took a preliminary dive into the new official baseball. We didn’t arrive at a concrete answer, but we did note that the distance on batted balls between 25 and 40 degrees had decreased. Guess whose barrels all fall within that range? With a different baseball, some of Carpenter’s barrels might have cleared the fence or sailed over the heads of outfielders, but alas, he picked the wrong time to experience a power surge.
What about his other, non-barreled hard-hit balls? One of them, a scorching 101.1 mph ground ball, resulted in one of his three hits. That’s good! The rest, however, are an assortment of less successful grounders and weaker versions of his fly ball barrels. There’s a lone line drive. Again, as with barrels, the story repeats itself – 63.6% are hard-hit by definition, but come with several flaws, whether they are insufficient exit velocities or awkward launch angles.
I’ve focused a ton on the negatives, but in the end, Carpenter’s weird production seems like it comes down to equal parts rotten luck and batted ball mishaps. With how often he’s striking out – his K% has been climbing since 2016 – it’s unlikely he’ll catch up to his .369 xwOBA. It’s also unlikely that he’ll remain at his actual wOBA of .171. As the sample size becomes more robust, they’ll meet somewhere in the middle. But maybe there’s a reason to lean optimistic. I borrowed a page from Devan Fink’s book and scraped data from 2019, the previous “regular” season, and looked at how well a hitter’s Barrel% from April correlated to his Barrel% across the entire season. The metric is a great descriptor and predictor of offensive success, and it’s Carpenter’s that stands out. Yes, it’ll go down, but maybe traces of it will remain:
Not bad! Hitters seem to retain way more of their early-season rate than I thought. Now, this is a heinous use of statistics that should never be replicated in a formal setting, but for fun, I plugged in Carpenter’s 2021 Barrel rate into the formula for the regression line. The result: 17.9%, which would represent the highest rate of his career in the Statcast era. There’s a near-zero chance he sustains a Trout-esque mark for an entire season, of course, but the bottom line is that April numbers aren’t totally meaningless. It’s downhill from there, but that’s not as bad as it sounds.
Overall, it’s a bit complicated. Matt Carpenter’s gaudy Statcast numbers are misleading due to his poor contact rates and inability to generate line drives thus far, which benefit him the most. Instead, myriad fly balls aren’t traveling like they used to, or don’t have enough speed behind them. At the same time, you can’t just ignore what Carpenter has accomplished. His shortcomings are still elite by major league standards. Ultimately, a sample of 22 batted balls just isn’t enough, and only time will tell whether the veteran slugger will follow a downward trend that began in 2019, resurge to his former level of play, or even unlock the next step in his career.
Justin is an undergraduate student at Washington University in St. Louis studying statistics and writing.
So Matt Carpenter is the exact same hitter he has been the last two years… not a buy-low candidate?
Ryan Zimmerman looked like toast few years ago before reharnessing the strike zone to go with his batted ball metrics, but it looks like Carpenter has completely gotten away from the doubles+ approach that made him viable.
For me personally, when it comes to 35 year old major league players, if the last 700 PAs tell me they are toast, I will believe they are toast until actual results begin to tell me otherwise. That’s now a full 162 games worth of PA, spread out over 3 year period, during which none of the actual results support anything other than, this guy is a guy who, for whatever reason, skills have collapsed enough that he is no longer a good hitter. I love x-metrics but at some point, and I guess that point is the intersection of the aging curve and 700 PAs of being not good, I’m siding with the actual results as the strongest indicator of what a player is/is not.
The thesis is “bad backs never go away, really” and if your guy has a bad back they better be falling from Clayton Kershaw territory and not Matt Carpenter territory.
The antithesis is “this guy’s x-stats weren’t good the last couple of years and he was mediocre, and now they’re decent and he was terrible” and that’s deserving of another shot somewhere (although likely not on the Cards, who have a clearly superior option at first base and the injuries will likely prevent him from doing anything else).
I think the synthesis is “even if you are willing to wait for the performance to regress back to the underlying batted ball data, that back is going to go out again” and you should probably not expect much and be pleasantly surprised if something nice happens.