The New Baseball Still Seems Juiced

Blake Snell, one of two 2020 playoff team aces the San Diego Padres acquired in offseason trades, has yet to allow his first run while pitching for his new employer. In five starts in spring training (which concluded on Tuesday), Snell yielded only five hits and never allowed an opponent to cross the plate, giving him the most scoreless innings pitched (14 1/3) and batters faced (52) of any pitcher in the Grapefruit or Cactus Leagues. Normally, it wouldn’t be noteworthy that a good pitcher shut down Double-A–quality competition over the equivalent of two strong regular-season starts. (Well, two strong starts for some pitchers; Snell hasn’t gone seven innings in an outing since April 2, 2019.) But because of a comment Snell made on March 9, his string of goose eggs seems somewhat significant.

“It’s definitely a different ball,” Snell said on a videoconference, comparing the current baseball to the models from past seasons. The southpaw explained that the laces on this spring’s ball were thicker, allowing him to get a better grip and more easily snap off breaking balls. He also asserted that balls in the air weren’t carrying quite as far. Two days earlier, another former Cy Young Award winner, Yankees ace Gerrit Cole, had said something similar, observing that the new ball’s seams seemed more comfortable and consistent. The righty would finish the spring with 24 strikeouts and a 2.45 ERA in five starts (and 18 1/3 innings) of his own.

The two high-profile pitchers weren’t imagining things: MLB really did deploy a new ball this spring, and that ball will be in action when the regular season starts on Thursday. Snell’s and Cole’s anecdotal but expert perceptions may have even nailed some salient aspects of the ball’s feel and effects, and the ball may help pitchers miss more bats (much to the dismay of fans—and league executives—who are hoping to avoid a 16th consecutive increase in the average strikeout rate). But despite that pair’s spring success, the early returns suggest that the new ball is far from dead. In some respects, it seems more lively than ever. In fact, batted balls this spring have been more likely to leave the yard than they were in any preceding exhibition season dating back at least 15 years.

As the graph below shows, the rate of home runs on contact—or home runs divided by batted balls, excluding sacrifices—slowly but fairly steadily increased over most of the century-long “live ball era” (to say nothing of the dead ball days before that). But the ball was never nearly as lively as it’s been since the second half of 2015, when a sudden spike in power brought the ball to baseball analysts’ attention. The home run rate kept climbing in succeeding seasons, and although statistical analysis and feedback from players made it look highly likely by 2016 and almost certain by 2017 that the ball was to blame, MLB mostly demurred, denied, or dissembled until 2018, when an MLB-commissioned committee made up of scientists and statisticians identified the ball’s decreased drag as the culprit. The home run rate reached its peak—so far—in 2019, propelled by a further reduction in drag (at least partly attributable to shorter seams) and changes in batter behavior as hitters adapted to the homer-happy (and home-run-reliant) conditions.

Throughout that several-season saga, MLB never acknowledged manipulating the ball intentionally, and the league was never shown to have done so. If anything, though, the apparent lack of a conspiracy was more confounding than intentional tampering would have been. The league seemed incapable of controlling or explaining the behavior of its most fundamental piece of equipment.

In July 2019, commissioner Rob Manfred said, “If we make a decision to change the baseball, you’re going to know about it before we change the baseball.” Sure enough, news surfaced in February that the ball would be different in 2021, although the revelation arrived via a leaked memo that was transmitted from the commissioner’s office to teams (which led to the rather roundabout sourcing of an MLB.com story about the new ball that cited an AP report about the private memo, instead of simply saying that the source was “literally us”).

The memo explained that sometime in 2019, official MLB baseball supplier Rawlings had finally reexamined its recipe. Rawlings, which MLB co-purchased in 2018, produced prototype balls at MLB’s behest from late 2019 to early 2020 that “loosened the tension of the first wool winding.” The change was an outgrowth of a recommendation by the committee, which had advised the league to take steps to restrict the variation of the balls by shrinking the range of acceptable specifications. The committee members pointed out that all else being equal, a ball at the upper legal limit for the ball’s COR (coefficient of restitution, or bounciness) would fly 36 feet farther than a ball at the lower legal limit. In theory, loosening the wool winding was a way to recenter the bounciness baseline by decreasing the COR and, by extension, the speed off the bat and the batted-ball distance.

Per the memo, the result based on testing at “MLB’s independent laboratory” at UMass-Lowell was believed to be twofold: that the weight of the ball would be reduced by a negligible amount (less than a tenth of an ounce), and that the ball’s COR and CCOR (cylindrical coefficient of restitution, a means of measuring the bounciness at game speeds) would be slightly suppressed, shifting the average to be “more in the middle of the current specification range.” The memo—which may have been sent in anticipation of an impending Sports Illustrated report about scientist, sleuth, and serial baseball dissector Meredith Wills’s contention that multiple models of the ball had been in use in 2020—said that MLB had instructed Rawlings not to use the new ball in games prior to 2021, in order to allow “sufficient opportunity to test the performance.”

The league hasn’t suggested that the new ball will dramatically reduce the home run rate. The memo estimated that the modification would reduce ball flight by 1 to 2 feet on fly balls hit 375 feet, and an analyst quoted in a report by The Athletic’s Ken Rosenthal and Eno Sarris estimated that the decreased distance would reduce home run rates by about 5 percent. But based on stats from spring training—both basic box-score results and Statcast tracking data about the ball’s flight—there’s no indication that the balls in use during the 2021 Grapefruit and Cactus League seasons were even slightly less likely to go over the wall than they had been in previous springs.

Although a recent ESPN report noted that runs and home runs per game are down this spring compared to the past few Grapefruit and Cactus League seasons, those basic stats don’t account for the fact that some games this spring have been abridged because of special mercy rules put in place to ease players into the strain of a six-month season in the wake of pandemic-suppressed workloads last year. Through March 13, teams were allowed to shorten games to five or seven innings if both managers agreed, and managers were also allowed to “roll” innings (or end them with fewer than three outs) as long as their pitcher had thrown at least 20 pitches after a completed plate appearance. Seven-inning games were permitted throughout the spring schedule.

If we compare how many homers were hit per batted ball, rather than per game, then we don’t have to worry about confounding factors caused by decreased contact or by MLB’s recent resemblance to Calvinball. The chart below displays spring-training and regular-season HR/Contact rates in each season since 2006 (the earliest spring data available via MLB’s public records).

The first important takeaway is that while spring-training home run rates on contact are generally a little lower than the corresponding regular-season rates—likely because of differences in quality of competition, weather, and other factors—the two rates tend to track each other closely: As regular-season home run rates have soared, spring-training rates have tended to follow (or lead). The correlation between the two over the past 15 years is .91, where 0 indicates no connection and 1 represents rates in perfect lockstep. The second takeaway is that the HR/Contact rate this spring is higher than it has been in any other spring on record. Spring dingers aren’t down; they’ve reached a record level, topping 5 percent of non-sacrifice batted balls. And in six out of the past seven seasons, the regular-season rate has moved in the same direction that the spring-training rate did.

One crucial caveat is that multiple models of ball are often in action during spring training, as teams use up leftover stock from the previous season (or seasons) and also integrate brand-new balls of the type that will be used in the regular season to come. Spring-training rates have proved predictive in the past despite that murky mélange of baseballs, but this year, for the first time, we can partially pierce the veil. Past exhibition seasons have generated little public information from MLB’s ball-tracking technologies. This season, though, Statcast is installed and producing publicly accessible data at spring-training parks (two of which host two teams apiece). That gives us a sample of roughly 43,000 pitches and almost 19,000 batted balls, robust data sets with which to calculate average exit speeds and derive estimated drag from the reduction in speed as a pitch makes its trip to home plate. A higher-drag ball encounters greater air resistance and loses slightly more speed as the pitch flies 55-ish feet (or someday, maybe more) from the pitcher’s hand to home.

If we graph the daily drag averages of all balls thrown this spring, we see what appears to be a bimodal distribution with two apparent peaks—in other words, possible evidence of two different balls with distinct drag coefficients. This split between balls is much more pronounced than it has been during the regular season, when only one kind of ball is supposed to be used. Dr. Alan Nathan, an expert on the physics of baseball who has chaired the MLB-commissioned committee’s investigations about the ball, says that he’s also seen a higher standard deviation in the daily drag averages than he has in any of the past few regular seasons, which could be indirect evidence of balls sampled from two distributions.

Knowing that there may be more than one model of the baseball in use this spring—one with higher drag, one with lower drag—doesn’t help us much unless we can figure out which one is old and which one is new. If we arrange the average drag coefficients for each day in chronological order, we find that as spring training has gone on, the drag has increased (which Nathan has noticed too).

On the whole, 51 percent of pitches thrown this spring behaved like higher-drag balls, but that figure rose from 45 percent in the first half of the schedule to 57 percent in the second half. It seems intuitive that teams would be more likely to use up the old balls first and then incorporate the new balls as supplies of the old ones dwindle and the start of the season approaches. As expected, front-office personnel from multiple teams with Statcast systems installed at their spring-training parks—including the one with the highest average drag—told us that they started out using holdover batches of balls but then transitioned mid-spring to using predominantly new ones.

It would be consistent with the data, then, if the new ball is the one with higher drag, which doesn’t seem to be a change that the league made on purpose. (Lowering the weight would raise the deceleration due to drag—lighter balls are more susceptible to air resistance—but a difference of less than a tenth of an ounce would only explain about half the observed increase in drag.) An MLB source says the league’s lab has conducted drag testing for informational purposes, but there’s still no official specification for drag, which seems like tempting fate, considering how unpredictable and significant the effects of even slight changes to that attribute can be, and how heavily drag contributed to the 2015-2017 and 2019 home-run-rate inflations. According to Nathan, the memo’s estimate of a 1-to-2-foot reduction in the distance of deep fly balls was based not on field tests, but on laboratory values plugged into an equation, a calculation that presumed that drag would stay the same.

All else being equal—there’s that phrase again—higher drag should translate to less carry and fewer home runs. Yet the higher-drag balls also have a higher home run rate on contact, because they have a substantially higher average exit speed (by about 0.8 miles per hour). That’s the opposite of what we would expect to see if their COR and CCOR were really reduced, though Jeff Kensrud of the Washington State University Sports Science Laboratory cautions that those traits are dependent on temperature, moisture content, and how the balls are being stored. (Ten teams will use humidors this year during the regular season, up from five last year, but teams haven’t used humidors in spring training.)

The higher-drag balls also seem to move more: Fastballs gain an extra 0.45 inches of vertical movement, and curveballs add 0.3 inches, relative to the lower-drag balls. In possibly related news, the higher-drag balls have a higher whiff-per-swing rate, by almost 3 percentage points. None of these findings conflicts with what Snell and Cole said: A difference in seams could account for improved grips, greater comfort, and more movement, as well as elevated drag, which could cause balls to carry less but possibly still fly farther if they’re leaping off the bat at higher speeds.

Exit speeds and strikeout rates rise with each successive season, so in trying to lower last year’s baselines, MLB is swimming against the tide. That’s a tall order for a ball that doesn’t seem much altered in a defense-friendly way. This spring, the strikeout rate has risen again, to a hair under 25 percent of plate appearances, which would be a record in the regular season, let alone the typically lower-strikeout spring.

Walks are up, also—which, when combined with the higher homer and strikeout rates, results in 38 percent of plate appearances ending with one of the three true outcomes and no ball put in play. That’s up from 26 to 27 percent in the not-too-distant early seasons of this spring sample. And strikeout and TTO rates are even more closely correlated with regular-season stats than home-run rates are (.98). It’s going to take more than tinkering to stop these runaway rates.

In a couple of weeks, we’ll have a (hopefully) more pristine sample of a single ball from which to draw conclusions about what the big leagues will look like this year. For now, we can’t know with certainty which balls were used by which teams on which days this spring, so some amount of educated guessing is inevitable. “Because we didn’t have a clear way to determine which ball was being used for each play of each game, I wouldn’t say we have something definitive that I trust,” one front-office person says, but he adds that he’s heard “a lot of skepticism that this ball is effectively de-juiced” in conversations with other teams. Another front-office source who questions the rigor and transparency of the testing, and the accuracy of the memo’s projection for the ball’s behavior, offers a rueful, “MLB is going to MLB.”

If we’ve learned anything about the mercurial baseball’s behavior over the past several years, it’s that its next fluctuation is tough to foresee. What we can say for sure is that the median hit distance this spring is 4 feet longer than it was during the 2020 regular season, and the median fly ball distance is 10 feet longer. The average exit velocity is up too, despite the profusion of less practiced and talented players. And most inarguable, batted balls are turning into homers more frequently than they have in any of the past 15 preludes to Opening Day. If the new ball was supposed to suppress homers, it hasn’t yet done its job. Maybe it’s spring training for redesigned baseballs, too.

All spring stats are current through Monday’s games. Thanks to Lucas Apostoleris of Baseball Prospectus for research assistance.