This article is coauthored by sabermetrician Mitchel Lichtman, who also conducted the research on which it is based.
Whether it’s Scooter Gennett going deep four times in one game, Chris Carter waving a one-handed homer, Marwin González surprising himself by clearing the fence with what he thought was a flyout, or a record seven grand slams on one day, the 2017 baseball season has supplied constant reminders that we’re watching a game featuring unprecedented power. This year’s MLB batters have hit 2,395 home runs, which puts them on pace to break the all-time single-season record of 5,693, set in 2000, by almost 350 homers.
This fusillade has left leaguewide totals looking a lot different from just a few years ago: Although we’re through less than 40 percent of the season, this year’s home run count has already exceeded 57 percent of 2014’s full-season total. The three-year increase in home runs per batted ball from 2014 to 2017 dwarfs the largest over any previous three-year span, including the notable increases that arose at the end of the dead-ball era and World War II, during the 1930 and 1987 "rabbit-ball" years, after the switch from Spalding to Rawlings as MLB’s baseball supplier in the late 1970s, and during the dawn of the PED era in the early-to-mid-1990s. We’ve never seen the home run rate soar anywhere close to this quickly without an answer other than "the batters got better" suggesting itself. Although the latest (and most extreme) spike’s origins are more murky, new data procured by The Ringer and presented below points back to the ball being involved again.
The current barrage began in earnest immediately after the 2015 All-Star break, leading to the largest increase in the rate of home runs/batted ball between a season’s first half and second half since at least 1950 (when Baseball Prospectus’s database begins). That increase coincided with a significant rise in the average exit velocity of batted balls, which largely explained the extra home runs but effectively replaced one mystery with another, leaving us wondering instead why balls were leaving bats with such speed. (Giant baseball-obliterator Aaron Judge didn’t debut until late last year, so we can’t blame him.)
The skyrocketing home run rate has served as a statistical sword in the stone for baseball reporters and analysts, who’ve scoured the available data in search of elusive answers. My own previous work with Rob Arthur of FiveThirtyEight, and the work of other writers, has examined and dismissed or discounted several theories about the home run surge, including hotter temperatures (which, in addition to being insufficient to explain the rising home run rate, wouldn’t affect batted-ball speed), a change in the strike zone, the optimization of batting orders, an influx of young home run hitters, teams shutting down top pitchers earlier in the year, hitters acclimating to faster pitches, a return to widespread PED use, pitchers aiming higher in the zone or throwing more meatballs, and pitchers throwing harder. (While pitchers are throwing harder, the speed of a batted ball is determined much more by the batter and the ball’s construction than the pitch speed. All else being equal, boosting pitch speed by 1 mile per hour increases batted-ball distance by less than 1 foot.)
It’s likely that the record home run rate has more than one contributing cause, so the single smoking gun that addresses every issue is probably a pipe dream. One proposed explanation that’s come to be known as the "air-ball revolution" — a purported data-driven trend toward hitters elevating their swings and batted balls — is particularly persuasive, meriting further mention later in this piece. However, coaches and players, conspiracy theorists, and analysts alike have kept returning to the ball, for a few reasons.
First, we know that changes in the ball have produced dramatic rises and reductions in offense before, not only in the big leagues but also in Japan, the NCAA, and the Mexican League. Second, the suddenness of the spike seems to downplay other possibilities: Because the wave crested so quickly, it’s hard to imagine a player-driven change (in either approach or pharmacology) producing so rapid a midseason rise. (The PED scenario would require widespread, almost simultaneous adoption of a highly effective drug that helps only hitters, isn’t detectable, and doesn’t make bodies look dramatically different.) But the home run pattern could be consistent with a new ball being introduced in the second half of 2015, when some teams tend to replenish their stocks. Third, Arthur and I showed last year that despite significant overlap in the Triple-A and MLB player pools, the home run explosion hadn’t yet extended to the upper levels of the minor leagues, which use balls manufactured in China that differ from the Costa Rica–made MLB ball. And fourth, the homers have been distributed disproportionately, flattening the distance between hitters on the home run leaderboard rather than inflating any individual totals to record highs. Less powerful hitters have derived bigger benefits, which seems consistent with the fact that there’s a fly-ball-distance sweet spot where balls become much more likely to turn into homers. Players who previously had warning-track power might have more to gain from adding extra feet to their flies and regularly reaching that sweet spot than the elite sluggers who were already comfortably clearing the fence.
The strongest argument against the "juiced ball" theory comes from MLB officials (including the commissioner), who claim that the ball hasn’t changed. Until recently, MLB offered no evidence to support its assertion. Last month, though, I relayed the contents of a previously unreleased document combining reports on two testing periods produced by MLB’s official ball-testing facility, the Baseball Research Center at the University of Massachusetts Lowell, which summarized the results of ball-testing conducted at the 2016 All-Star break and in February 2017. The 2016 report concluded that: "There is no evidence from the results of this study that the performance of the 2016 regular-season baseballs … would have resulted in any difference in on-field performance from those used during recent seasons." Additionally, Alan Nathan, an expert on the physics of baseball whom MLB paid to provide his own analysis of the 2016 test results, told me, "I saw nothing in the data that was presented that suggests that the ball has been altered at all." (Nathan and an MLB spokesman declined to comment on this article prior to publication.)
Although the source of that MLB-ordered report justified some skepticism, its findings appeared to set some of the speculation about the ball to rest. But without an equally convincing alternative, we couldn’t declare the baseball case closed. Now, new evidence has arisen that seems to support a contradictory conclusion: that much of the rise in home runs can be explained by the ball.
The concept of a "juiced ball" refers mainly to the ball’s coefficient of restitution, or COR — basically, its bounciness. The higher the COR, the faster and farther the ball travels after it’s hit by a bat. But COR isn’t the only factor contributing to a batted ball’s distance. The ball’s circumference and seam height matter, too. While the latter two factors don’t significantly influence exit speed, the smaller the ball and the lower its seams, the greater its potential to carry. A 2013 study sponsored by the NCAA found that lowering the seam height from the NCAA balls’ then-standard .048 inches to .031 inches (thereby reducing the drag effect of air resistance) made a ball with the speed and trajectory of a typical home run fly 20 feet farther on average. (The higher the exit speed, the more mileage the lowered seams added.) When the NCAA began using the flatter-seam ball in 2015, the home run rate rose by 44 percent relative to 2014 — as it happens, almost the same increase we’ve seen in the major league rate from 2014 to 2017.
Mitchel Lichtman, my coauthor on this article, is a former consultant to MLB teams, the creator of ultimate zone rating, and the coauthor of The Book: Playing the Percentages in Baseball. Late last year, Lichtman commissioned independent ball-testing in an effort to confirm or refute the altered-ball hypothesis. First, he purchased 36 game-used MLB balls from eBay, each of which was authenticated for a particular game with an official MLB collector’s holographic sticker. Seventeen of the balls were used in games played prior to the 2015 All-Star break, ranging from May 2014 to July 2015. Nine were used in games in either August or September 2015, and the remaining 10 were used in May, June, or July 2016. The balls seemed to be in good condition, and the three groups were indistinguishable from each other by outward appearance and feel.
Lichtman sent the balls to one of the few facilities capable of rigorous testing, the Sports Science Lab at Washington State University. There, the balls’ CORs were tested by firing them at 120 mph into a steel cylinder, six times each, which is considered the closest approximation of in-game collisions that wouldn’t destroy the ball. The lab also measured the circumference and weight of the balls, the height of their seams, and their dynamic stiffness (a more sciencey way to say "hardness"). Before testing, all of the balls were stored for two weeks in a humidor with constant temperature and humidity to ensure consistent conditions, and randomly coded so that no one at the lab knew which balls came from each lot. (Lichtman kept the key that matched codes to baseballs.)
The testing revealed significant differences in balls used after the 2015 All-Star break in each of the components that could affect the flight of the ball, in the directions we would have expected based on the massive hike in home run rate. While none of these attributes in isolation could explain the increase in home runs that we saw in the summer of 2015, in combination, they can.
Although the results did reveal differences in all of the relevant categories when Lichtman compared the balls used before the 2015 All-Star break with all of the balls used after, there was no way to determine whether the balls he sampled from the second half of 2015 were newly manufactured or holdovers from the first half. To avoid that complication, he compared the 17 balls from before the 2015 All-Star break with the 10 balls from 2016, although the full results for every ball tested are available here.
The table below displays the average differences between the old and newer balls, as well as the estimated resulting increases in distance and actual or equivalent exit speed, according to Alan Nathan’s trajectory calculator and the aforementioned NCAA seam-height study.
The newer balls have higher CORs and lower circumferences and seam heights, which would be estimated to add an average of 7.1 feet to their distance, equivalent to the effect we would expect to stem from a 1.43 mph difference in exit speed. Although those differences don’t sound enormous, Nathan has noted that "a tiny change in exit speed can lead to much larger changes in the number of home runs." Last July, he calculated that an exit-speed increase of 1.5 mph would be sufficient to explain the rise in home runs to that point, which means that the 1.43 mph effective difference that Lichtman’s analysis uncovered could comport almost exactly with the initial increase in home runs. Lichtman calculates that a COR increase of this size, in this sample, falls 2.6 standard deviations from the mean, which means that it’s extremely unlikely to have happened by chance.
With the newer balls’ reduction in circumference comes a decrease in weight, although according to Robert Adair’s book The Physics of Baseball, the ball’s weight, independent of its other qualities, has little effect on flight distance. Similarly, while dynamic stiffness does affect the flight of balls hit by the hollow bats used in amateur ball, it doesn’t play a role with the solid bats used in the big leagues. However, a dynamic-stiffness difference that large does add to the evidence of altered composition.
There’s no indication that any of the balls Lichtman had tested fall outside of MLB’s allowable ranges, but some of those ranges are laughably large, leaving a lot of leeway for legal variation with major effects on the field. As an earlier ball-testing report by the Baseball Research Center that was publicly released in 2000 acknowledged, "two baseballs could meet MLB specifications for construction but one ball could be theoretically hit 49.1 feet further." According to Nathan, an increase in COR of .012 could completely account for the 2015-to-2016 home run revival, which makes the league’s allowable COR range of .514 to .578 for BRC’s standard flat-surface COR test (which reports higher COR values, for the same balls, than the cylindrical-surface test that Lichtman commissioned) seem absurdly imprecise. We should note here that since seam height and circumference, unlike COR, don’t directly affect actual exit speed, these ball changes can’t explain the entirety of the Statcast-derived exit-speed increase from the first half of 2015 to 2016. We can’t account for that discrepancy, although inconsistencies in the always-evolving Statcast system could be responsible for some of it.
Data issues complicate matters in one other way. In theory, we should be able to tell whether seam height and circumference are causing balls to fly farther by comparing the distances traveled by less and more recent batted balls with the same exit speeds and launch angles, but large-scale changes in Statcast- and stringer-reported data make it difficult to compare distances between 2015 and 2016. However, an analysis by Arthur that will be published this week at FiveThirtyEight does show that even after accounting for exit velocity and launch angle, balls hit in the first half of 2016 produced a higher home-run-per-fly-ball rate than balls hit in the first half of 2015, which supports the suspicion that the 2016 balls carried farther. "Balls hit with the same exit velocities and launch angles were much more likely to become home runs in 2016 than in 2015, suggesting that their air resistance might have decreased," Arthur told us via email.
Although our analysis contradicts the "nothing to see here" conclusion of the MLB-funded reports, it doesn’t contradict all of the data that those reports present. MLB still hasn’t released the full results of the Baseball Research Center’s testing, and what we do have lacks historical results for the superior cylindrical COR test that the Sports Science Lab performed for Lichtman, instead displaying previous values for only a lower-speed (and potentially less accurate) COR test in which the ball is fired against a flat surface.
Depending on one’s interpretation, the summary data displayed in graphical form within the BRC report that MLB provided actually aligns fairly well with the results of Lichtman’s testing. The BRC-reported COR values for 2016–17 are noticeably higher than they are for 2014, and the circumference and (especially) seam height are smaller and lower, respectively (albeit bigger/higher than at previous points in the sample). Finally, the latest results reveal that the baseballs haven’t been harder since testing started in 2004.
Using a graphics program to overlay grids on the BRC graphs, Lichtman estimated the values of each data point in the BRC report. His measurements of those values yield COR, seam height, and circumference differences from 2014 and 2016 that he concludes (again via Nathan’s trajectory calculator) can explain 72 percent of the home-run-rate change between the two seasons. That percentage does shrink substantially if, instead of solely comparing 2014 with 2016–17, he includes the COR results from BRC’s first test of 2015, which surprisingly yielded the highest figure of the past few seasons. However, along with the testing that Lichtman arranged and the testing that Arthur and I commissioned for FiveThirtyEight on balls purchased from Rawlings in late 2015, three separate sources have reported climbing COR averages of some size over a period when the home run rate also increased.
Although the change in MLB’s home run rate from the first half to the second half of 2015 was much larger than the subsequent changes from the second half of 2015 to 2016 or from 2016 to 2017, it has stayed on a steady upswing even after the initial surge. That brings us back to the air-ball revolution. Although home-run-rate inquiries are often framed as either/or debates — has the ball changed or have swings changed — the best answer is probably "both," much as it was when Moneyball misleadingly framed baseball’s quest for intelligence as an adversarial "stats vs. scouts" proposition.
We know that some hitters in recent seasons have intentionally tried to hit more balls in the air, and several of them have made themselves much more dangerous at the plate. The idea of elevating the ball isn’t new, but in the past year or two it seems to have made major headway in combating the common belief that it’s better to hit down on the ball. That could be because hitters are trying to avoid increasingly common infield shifts, or it could be because Statcast has made it easier to identify hitters who could benefit from lifting their launch angles. (MLB teams have had access to exit velocity and launch-angle data since at least 2009, but Statcast’s high profile since its full rollout in 2015 has raised public and player awareness of those metrics’ existence and value.)
Maybe, though, it’s because fly balls bring richer rewards than they did in 2014. If the ball is capable of carrying farther today than it was a few years ago, it makes more sense to try to let it fly. Evidence of an altered ball doesn’t invalidate the idea that hitters have something to do with this season’s unparalleled power display. It simply suggests that players are paying attention. And not only in the majors: We haven’t tested any minor league baseballs, but the Triple-A home run rate has also seen a huge spike since the middle of last year.
There’s precedent for this sort of call-and-response relationship between changing balls and batters. As historian and sabermetrician Craig Wright recently detailed in his newsletter, Pages From Baseball’s Past, the shift from the dead-ball era to the live-ball era, which is usually traced to 1920, actually unfolded over a three-season period from 1919–21. In 1919, with the price of American wool inflated by a World War I–caused shortage, the Reach Company, which then manufactured the MLB ball, began importing Australian wool to use as its yarn. Australian wool, which came from Merino sheep, was springier than the American kind, which made the new balls bouncier. As the 1919 season progressed and the "Merino ball" replaced the leftovers, the leaguewide slugging percentage rose.
Offense increased again in 1920, as MLB began replacing used baseballs with fresh ones much more frequently within games, largely in an effort to police a new partial ban on doctoring the ball. Because the balls were in better condition, they were easier to drive. The following year brought another rise in run-scoring, this time not just because the ball was replaced even more frequently, but because hitters looked around and realized that the new game deserved (literally) loftier goals than just putting the ball in play. As Wright recounts, "The biggest factor in the big offensive leap in 1921 was traceable to a change in the approach of some hitters — they begin to shift from a general goal of hitting liners and hard grounders to driving more balls in the air." Led by Babe Ruth, batters adapted to a bouncier, better-preserved, and (as of an earlier change from the 1910s) lower-seamed ball and started swinging for the fences. It all sounds familiar, and today’s bigger, stronger, less strikeout-averse hitters almost certainly swing harder than ever before.
Even if MLB’s peak-power era began with a change in the baseball, that wouldn’t imply a plot at the highest levels of the league. Yes, a Ken Rosenthal report from January 2015 revealed that MLB had brought up the idea of juicing the ball, and yes, it does seem somewhat suspicious that the ball began flying just when fans and officials were fretting about run-scoring falling to its lowest level in almost 40 years. But modest alterations to multiple components of the ball could have gone undetected more easily than a massive alteration to one aspect. The change could have been an unintended consequence, just as it was when the Reach Company tried to cut costs on wool and ended up with a bouncier baseball. Even MLB’s testing and allowable range suggest that the ball varies slightly from batch to batch and year to year, and those slight differences might stack up in certain seasons.
Nor would an altered ball be a bad thing by default. It’s true that a juiced/flat-seamed baseball could encourage a "swing-for-the fences" approach that results in more strikeouts and fewer balls in play. (MLB’s K rate actually decreased in the first half of 2015, relative to its 2014 level, but the home run surge has coincided with a renewed rise in strikeouts.) It’s also true that total transparency about the ball seems like something worth wishing for. But many fans may prefer baseball’s current scoring and home run rates to the lower levels that predated the dinger surge, however we’ve arrived at this point. And maybe we’re better off with a bona fide baseball mystery than another easy answer. In an era that seems to supply a statistical response to almost every question, it comes as a kind of relief that there’s still something so central to the game that we can’t quite figure out.
The technology and language we use to dissect and discuss scoring changes in baseball look and sound a lot different than they did in Ruth’s day. After weighing all the evidence from our high-speed cylindrical COR tests and cameras, computers, and radar, though, the conclusion suggests that history is probably repeating itself. As was the case almost a century ago, it’s likely that both the ball and the batters have changed. And just as my millennial mind explained how the dead ball came to life, 22nd-century fans might look back on the long-ball bonanza of the two-thousand-teens and believe that the baseball started it.