Has MLB Pulled a Switcheroo with the Baseballs This October?

For a moment, it looked like Will Smith would be the hero. In the bottom of the ninth, sandwiched between the two cataclysmic half-innings that abruptly ended the 106-win Dodgers’ season, they had a brief flicker of hope when with one out and one on, Smith hit a drive off Daniel Hudson that looked as though it might — might — make him the hero, with a walk-off home run that sent the Dodgers to the NLCS. It was hardly implausible given that the 24-year-old rookie had hit two of the Dodgers’ major leauge-leading seven walk-off home runs this year, or that nearly half the drives hit to the specifications of which he struck Hudson’s hanging slider — 100.3 mph, at a 26 degree launch angle — have left the yard over the past five seasons.

It wasn’t to be.

Smith’s drive fell short as, ultimately and in more gruesome fashion, did the Dodgers. There will be plenty of time to dissect the larger situation but for the moment, consider the batted ball, which had a 69% of becoming a hit and a 46.1% chance of going out based on similarly struck spheroids. When it didn’t, it was just the latest in the genre of hold-your-breath moments that wound up producing mutterings that maybe the baseball has been de-juiced this October — that is, that the postseason ball is different from what’s been used in the regular season.

It’s not hard to understand why this notion has taken hold. So far this month, we’ve seen home runs hit at a lower frequency than during a regular season that set all kinds of records for long balls, and scoring rates have fallen as well. In the blur of Division Series games, many a hard-hit ball appeared bound to go out — at least based upon the way our brains have become calibrated to this year’s nearly-numbing frequency — only to die at the warning track. Yet it’s harder to make the case that something is different given a closer look at the numbers, both traditional and Statcast, at least if you’re not Baseball Prospectus’ Rob Arthur, whose model to calculate the drag on the baseball by measuring a pitch’s loss of speed does suggest something is afoot. More on his latest findings below, after I present my own analysis.

…

Fueled primarily by more aerodynamic baseballs, with smoother leather and seams so low that the balls are nearly round, batters hit a record 6,776 homers during the regular season, 671 more than the previous high set in 2017. On a per-game basis, the 1.39 homers per team was 21.4% higher than last year, 11.0% higher than the record set in 2017, and 61.9% higher than in 2014, the year homers reached their post-strike low point (0.86 per team per game). On a per plate appearance basis, this year’s 3.63% rate was 20.4% higher than last year, 10.3% higher than in 2017, and 59.6% higher than in 2014, while on a per batted ball basis, the 5.42% rate was 21.8% higher than last year, 12.5% higher than 2017, and 67.7% higher than in 2014. You get it.

Any way you slice it, it was a lot of homers, too many for some tastes, my own included. So far in October, we’re seeing fewer of them, though they’re hardly scarce. Through Wednesday’s games, teams averaged 1.29 homers per game, a drop of 7.6% from the regular season, but a rate that’s still higher than the 2017 regular season mark. For what it’s worth, per-game scoring has dropped by 13.4%, from 4.83 runs per game in the regular season to 4.18.

Both drops are in line with Wild Card-era history. As YES researcher James Smyth noted back in March, from 1995-2018, playoff teams’ per-game scoring fell 16% relative to the regular season, but their per-game home run rate fell by just 8%. Better pitching, better defense, and colder weather have an impact upon both rates, but moreso on scoring. For what it’s worth, on a per plate appearance basis, this October’s 3.41% is about 6% below the regular season, but the per batted ball rate of 5.47% is up a hair, from 5.42%! The rise in strikeout rates, from 23.0% during the regular season, to 27.0% in the postseason, is playing a significant part in the per-game homer decrease.

So this all may be just in our heads, part of the casual chatter in press boxes, broadcast booths, and on social media. Watching various Division Series games, a few drives come to mind as ones that tripped my own switch — and sometimes those of the play-by-play voices — as far as believing a home run was in progress. As I wrote a couple weeks ago, often, it’s steep launch angles that produce such a response, since exit velocity is harder to determine in that split second that the camera is focused on the batter, and at that point we don’t yet know the distance.

Here are clips of a few drives that caught my eye, all from Monday’s four-game smorgasbord, First is the Braves’ Adeiny Hechavarría, batting against the Cardinals’ Andrew Miller with the bases loaded in the top of the seventh inning:

That’s 94.1 mph off the bat with a 32.5 degree launch angle, a combination that only produces a homer between 7.5 and 10.5% of the time (I’m using 32 and 33 degrees with the Statcast Exit Velocity & Launch Angle Field Breakdown). It sure looked more impressive to start, perhaps in part because of the tension of the situation and the unlikelihood of the batter’s heroics. The Cardinals led 4-3 at the time, but the Braves were trying to cash in on a sun-aided leadoff triple by Ronald Acuña Jr.; Hechavarría homered a career high nine times this year in 221 PA but produced just an 87.5 mph average exit velocity overall; he doesn’t usually put a charge into one like that. A bit more oomph and/or a bit closer to the foul line and that one had a chance to go out, but as it was, it was a fairly routine fly ball.

Here’s the Twins’ Marwin Gonzalez, representing the tying run against the Yankees’ Chad Green in the sixth inning of Game 3 of that series:

That’s 100.2 mph off the bat with a 37.9 degree launch angle, a bit steep; that combination produces a homer 21.5% of the time, a rate about half of what it would be at the same speed but with a 33-degree angle (46.3%). This one went an estimated 359 feet, not quite enough to get it over the wall marked at 365 feet.

Those weren’t quite the close calls that they initially appeared to be. This one, off the bat of the Dodgers’ Max Muncy — the subject that keyed my recent look at launch angle porn — against the Nationals’ Sean Doolittle, is almost always a home run:

That’s 107.1 mph off the bat, with a 31.9 degree launch angle, a combination that produces a homer 94.5% of the time. Statcast’s 400-foot estimate might be generous by a couple feet given where it was caught relative to the 402 sign, but regardless, Muncy did not maximize the impact of that drive because he hit to the fat part of the ballpark, and had the wind working against him. “I think I might have had a little bit of help there from Mother Nature,” said Doolittle after the game. “I think we kicked up the wind a little bit,” said catcher Kurt Suzuki.

Here it’s worth pointing out that those home run percentages are based on five years worth of outcomes in all ballparks. The presentation is overlaid on a map of Cleveland’s Progressive Field, but its dimensions don’t factor into the percentages. It’s also worth noting that none of these percentages account for the spin of the ball, which produces both lift and drag as it increases. Here’s FanGraphs alum Travis Sawchik quoting the game’s foremost physicist from a September 12 article:

Alan Nathan, physics professor emeritus at the University of Illinois and MLB consultant, says there are optimum spin levels in baseball for batters. “Not only does the lift increase with the spin, but it’s really only been in the last few years that we’ve realized that the air drag also increases with the spin,” Nathan told FiveThirtyEight. “So that slows the ball down.”

In that piece, Sawchik noted that spin rates on batted balls this year have declined, producing less drag, a potential factor in this year’s home run surge. For what it’s worth, all three of the drives I illustrated had spin rates below the median of the respective launch angle buckets shown in Sawchik’s article (2,751 rpm for Hechavarria’s drive, 2,484 rpm for Gonzalez’s, 2,264 rpm for Muncy’s), suggesting at the very least that their spins weren’t hindering the ball’s flight:

“Certainly by 2,500 rpm it starts to level off, meaning with additional spin you don’t get any increase in distance [as a hitter],” Nathan said. “Then if you go even higher, maybe above 3,000 or 3,500 rpm, that’s the [range] where distance actually starts to decrease with spin as the drag starts to overwhelm the good benefits that you get with the additional lift.”

Meanwhile, it’s worth noting that the average fly ball distance this October is higher than it is in the regular season, a quirk that we haven’t seen since 2015, and the expected wOBA on fly balls is higher:

Fly ball distances through Wednesday have increased by an average of four feet, and xwOBAs on flies is up by 70 points. Both of those differences run counter to what we’d expect if the balls being used are less lively than ones in the regular season. Those longer fly balls and higher xwOBAs are the result of balls being hit harder, an average of 93.5 mph, compared to 92.0 in the regular season.

Given that it’s October, it’s of course worth noting that all of this is happening despite cooler temperatures, which will cut into a fly ball’s distance. In a 2017 paper, “Effect of Temperature on Home Run Production,” Dr. Nathan calculated that for temperatures between 50 and 100 degrees Fahrenheit, fly ball distance increases (or decreases) by 0.33 feet for every degree the temperature rises (or falls). We don’t know what the temperature was at the time those balls were hit — MLB only publishes the first-pitch temperature — but, for example, the starting temperature of the Yankees-Twins game at Target Field was 62 degrees, 18 degrees cooler than the starting temperature on September 21, the Twins’ second-to-last regular season home game; under those conditions, a fly ball in Game 3 would travel six feet fewer than the earlier one when struck under identical conditions. It’s not hard at all to imagine a drop of 10 degrees from first pitch to the bottom of the ninth; that would cost the batter 3.3 feet. For the Dodgers and Smith, that might have been enough to avoid a crushing defeat.

…

All of that certainly leaves room for a deeper analysis by somebody more adept than myself when it comes to the physics of lift, drag, and spin such as Arthur, who was answering the Bat-Signal even as I was putting this together. On April 5, with just six days of 2019 data underfoot, he had noted, “the drag numbers so far are among the lowest recorded in the last calendar year” and concluded that “the baseball has played like it’s much more aerodynamic than it was last year. In fact, the current drag coefficient is really only comparable to 2017, when the baseballs were more aerodynamic than they had been in at least a decade.” Arthur acknowledged that drag had a tendency to fluctuate over the course of the year, based on “natural manufacturing variation or differences in the underlying supplies of cowhide and thread,” but believed that we were in for a big year, homer-wise, which proved to be the case.

In his latest analysis, Arthur notes a drastic change:

In the last week’s worth of Division Series games, the drag coefficient spiked to a high it hadn’t regularly sat at since 2016. October days only contain a fraction of the games of a typical regular season night, but we’re still dealing with a sample of more than 800 fastballs to measure drag with. The probability that a random selection of games from the rest of the regular season would feature as much air resistance as we’ve seen so far in the postseason is about one in one thousand. This was an abrupt spike, as well: It’s the largest change in drag coefficient from week to week this season, by a factor of three.

Arthur also built a model of home run probability based on exit velocity, launch angle, and a park effect, which he deployed in August. Using it on October data:

In total, this model predicted that there should have been 24 more home runs so far in the playoffs than the 43 that were actually hit–an astounding 50 percent difference. For example, Muncy’s monster blast had a 95 percent probability of clearing the fence based on the regular season baseball. Instead, it fell several feet short. Two other fly balls had in excess of a 98 percent chance of being home runs before becoming outs instead.

…The model predicted that around 30 batted balls had between a 20 and 50 percent chance of becoming home runs, and that about 10 of them should have made it. Instead, only one of that group so far this postseason has landed beyond the fences

Arthur’s work squares with the wide perception that things have changed, leading him to conclude:

Almost overnight within the same season, the ball has been replaced by one with wholly distinct characteristics, ranging from the speed with which it leaves the bat to the distance it travels. It’s hard to believe that MLB, which owns the company that produces the baseballs, would consciously approve a change to their manufacturing on the eve of the most important month of the season, but it’s also hard to believe that this would happen by accident.

Even so, Arthur’s findings don’t explain the longer fly ball distances I noted, or the more or less equivalent rate of home runs on contact relative to the regular season. It’s certainly plausible that my findings are artifacts of a small sample size, but they’re there nonetheless, as a reminder that explanations are not always neat and tidy.

For what it’s worth, via Statcast data through games of March 28 (202 balls, the closest end-of-day total I can get to the current sample of 229 in this postseason), the average fly ball distance was at 325 feet, though by April 4, the day before Arthur’s piece ran, it was down to 317 feet, a fluctuation that suggests the measure hadn’t stabilized. Last year, in games through October 16, the 233 fly balls to that point averaged 321 feet, three feet longer than the final average, which was based on 363 fly balls. In 2017, through games of October 14, the 238 fly balls to that point averaged 312 feet, three feet shorter than the final average, which was based on 417 fly balls. So yes, things can change.

There’s still room for more analysis, not just with regards to larger sample sizes but also verification of physical differences in the postseason baseball relative to the regular season one (paging Dr. Meredith Wills and those who can supply her with postseason baseballs). What’s more, we have yet to hear from pitchers such as Justin Verlander telling us that the baseballs feel different, or have different seam heights — but then, if the current conditions are working to their advantage, if the ball has been deadened somehow and is producing more drag, they’re probably less likely to speak up than when they’re getting bombarded by homers. One way or another, particularly so long as this postseason runs, this discussion isn’t going away.