On Monday, Major League Baseball enacted a new rule that’s simultaneously subtle and profound. In the short term, it should have little effect on the field. In the long term, it may provide a template for fixing most of the aesthetic and stylistic problems facing the sport. It’s a potential panacea that you might not notice if you haven’t read the rule.
In deference to this season’s lockout-compressed spring training, MLB teams were allowed to carry 28 active players through May 1, but as of Monday the maximum is 26. That’s not the new part; rosters were capped at 26 players last season, too. (In 2020, when the active roster size was first slated to expand from 25 to 26, teams were ultimately permitted to carry 28 players throughout the pandemic-shortened 60-game season.) No, the new part is this: Starting today, teams are allowed to roster no more than 14 pitchers.
This restriction is both conservative and groundbreaking—one small step for each team, one giant leap for the league. On one hand, 14 pitchers is plenty. That’s enough for a five-starter rotation and nine relievers, or six starters and eight bullpen arms. A ceiling so high can barely be called a limit; most teams probably wouldn’t regularly roster more pitchers than that even if they could. On the other hand, any fine print about positional constraints is unprecedented: Roster sizes have fluctuated from century to century and decade to decade, but at no point have league rules imposed any position-based checks on how teams could apportion the space on their roster. And once in place, the 14-pitcher limit can and will lead to less lenient caps. Barring further postponements, the limit will drop to 13 four weeks later, on May 30, which will actualize a long-delayed provision originally scheduled to take effect in 2020. And the limbo bar could—and, I’ll argue, should—be lowered from there. When USA Today’s Bob Nightengale reported last May on the league’s planned pitcher limit, he noted that MLB “has hopes of further reducing it in future years”—potentially to 12 pitchers, or “maybe even 11 in ensuing years.” It’s not a new idea, but it’s one whose time has come.
Relative to leagues like the NFL and NBA, MLB has tended to take a laissez-faire approach to tweaking its rules, a byproduct of a powerful players union and an older audience that values tradition and statistical consistency. In recent years, though, that noninterventionist stance has started to change, as growing discontent with lengthening games, one-dimensional offense, and rapidly evolving (or devolving) pitcher usage, coupled with increased competition from other entertainment options, has prompted the league to test a wide assortment of experimental rules in independent partner leagues and the affiliated minors. We’re on the verge of a period of drastic on-field intervention, by MLB standards, one that could make pitch clocks, shift bans, automated strike zones, and other previously inconceivable developments part of the sport at its highest level.
If I had to identify a single path to improving what ails MLB’s on-field product—one that would do the most good with the least downside risk, and potentially obviate the need for other, more meddlesome measures—it would be phasing in more stringent restrictions on the number of pitchers teams can carry. Lowering the allowable number of active pitchers to 12, or eventually, 11—sorry, A-Rod, but 10 might be taking it too far—would elegantly target multiple problems, including decreased contact and scoring; less spectator-pleasing pitcher-usage patterns; long, slow-paced games; and rampant pitcher injuries. There’s almost no on-field bummer about baseball that this single, decisive remedy wouldn’t at least nudge in a more fan-friendly direction. And unlike many other measures MLB seems to be stumping for, stricter pitcher limits would do nothing to directly dictate or prohibit player and team behavior after the games begin. There was a time when I believed that limiting the proliferation of pitchers in this way was a step too far. I’ve come to believe that it can’t happen soon enough. Maybe I can convince you, too.
One of the stories of this season so far is an offensive outage that’s been blamed largely on MLB’s continued tampering with the ball. Despite the demise of pitcher hitting, stricter enforcement of the foreign-substance ban, the inflationary effects of the extra-innings zombie runner, and a lockout-delayed start to the regular season that bypassed April’s first (and, theoretically, coldest and most offense-suppressing) week, the league as a whole hit .231/.306/.369 with a .282 BABIP, a .675 OPS, and a 23 percent strikeout rate, and scored 4.12 runs per team per nine innings during the month. That’s the lowest batting average in April since 1968’s .230, the lowest OPS in April since 1981 (.674, and before that, 1972’s pre-DH .666), and the lowest BABIP, slugging percentage, and runs-per-nine total in April since 1992 (.280, .368, and 4.12), accompanied by the second-highest position-player strikeout rate ever (after last season’s 23.8 percent).
The graph above charts the level of offense in April in each season of the DH era (save for strike-delayed 1995 and COVID-delayed 2020). This year, the ball seems less lively than it’s been since the juicing that started in 2015. Whether or not MLB is putting multiple models of baseballs in play, as it secretly did last season—the league says it isn’t, but some players suspect it still is—balls in the air aren’t flying as far as they have for the past several seasons, thanks to some combination of increased drag and the installation of humidors in all 30 parks (up from 10 last year). Like last year, the lost homers have mostly turned into outs. However, while the ball has been deadened compared to the models in use during the homer-happiest seasons, it’s hardly flatlining. In fact, as the graph below shows, the rate of home runs per ball put in play is significantly higher than the historical average in the live ball era, slotting in between the wild card era average and the DH era average.
If anything, the new ball is just right, occupying a Goldilocks zone between the post-expansion lows of the early 1970s and highs of the late 2010s. The ball has borne the blame for a more sweeping problem, one that can’t be corrected through slight adjustments to the baseball’s bouncy center or seams. MLB made the problem more acute and unignorable by partly blunting the one weapon that was helping hitters keep up appearances, but relying on a ball that turned warning-track outs into no-doubt dingers wasn’t a solution either. MLB hoped that making long balls less likely would incentivize batters to swing for the fences less often and prioritize putting balls in play, but when pitchers are routinely pumpin’ cheese at a hundo and harnessing new technology to spin the nastiest breaking balls ever seen, and defenders are placed in the paths of what once would’ve been base hits, batters have it hard whatever they do. The fault in MLB’s offense doesn’t lie with self-defeating behavior on the run-scoring side of the equation, but with opponent-defeating advances in run prevention. Offense will rebound a bit as the weather warms, but not enough to overcome the fact that pitchers—and, to a lesser degree, defenders—have gotten too good, courtesy of analytically optimized training, pitcher deployment, and defensive positioning (in the outfield even more so than the infield).
I could dump a few more graphs on you here, but instead I’ll summarize a few effects. Starting pitchers have faced an opponent for at least the third time in a game in only 14.9 percent of their plate appearances, easily the lowest rate ever (and roughly half of what it was 20 years ago). Pitchers’ average four-seam fastball velocity continues to climb, to a new high of 93.9 mph. For the first time on record, non-fastballs—which are being thrown harder, too—make up the majority (51.2 percent) of pitches, as hurlers throw their bendiest, whiffiest stuff without regard for outmoded conventions of being behind or ahead in the count. According to analyst Cameron Grove’s Stuff+ model, which evaluates pitcher repertoires based solely on speed, movement, and release point, the resulting leaguewide uptick in stuff has produced measurable improvements in expected ERA, whiff rate, groundball rate, and hard-hit rate relative to 2015, the first year for which the metric is available. (This table compares April 2022 to previous full seasons, but April-only stats tell a similar story.) Most of the gains are attributable to more and better off-speed stuff.
Stuff-Driven Gains in Expected Results Relative to 2015
Stuff has gotten so good that it’s become a viable strategy for catchers to set up over the middle and for pitchers to put it right there. There’s no way to outlaw the strength programs that have helped pitchers throw harder, or to take away the high-speed cameras and tracking devices that have turned pitch design into a science, or to make pitchers unlearn what they’ve learned about concepts such as seam-shifted wake, spin mirroring, or vertical approach angle, or to make managers forget that starting pitchers get less effective the deeper they go into games. What MLB can do is make it harder for pitchers and teams to weaponize those insights, by reducing the size of staffs and closing loopholes that could enable teams to skirt roster restrictions. (Crucially, this week’s institution of a 14-pitcher limit coincides with the return of the 15-day non-COVID injured list for pitchers—up from 10 days—and another new rule that prevents teams from optioning a player to the minors more than five times per season, both of which should help curb roster churn.)
It’s common knowledge that pitching staffs have hugely expanded, but the dramatic reshaping of rosters is still striking when you see it laid out. We don’t have historical data on day-by-day roster composition, but we can approximate it by classifying the players who appeared for each team in each season over each span of seven games. The following graph shows the number of unique pitchers and non-pitchers per team used in each season over the average seven-game span, from 1901 to 2022. Both pitcher and non-pitcher counts have climbed over time as rosters have expanded and players have filled more specialized roles, but the more important takeaway is the steep ascent of the ratio of pitchers to non-pitchers used. (The combined counts of pitchers and non-pitchers per seven-game span slightly exceed the normal maximum roster size in each season, for a few reasons: Teams regularly reshuffle their rosters throughout the year; some of the seven-game spans came after late-season roster expansions; and players who appeared as pitchers and non-pitchers within each span were counted twice. There haven’t been enough two-way players or position-player pitchers for this to make much of a difference; a version of the graph that strips out pitchers who also made non-pitching appearances within the same span looks essentially the same.)
Up until right around the time that the Red Sox were realizing Babe Ruth might be an even better hitter than he was a pitcher, teams used about twice as many non-pitchers as pitchers in the average seven-game span. But the gap has gradually—and, over the past few decades, not so gradually—narrowed. Since 2017, teams have used more pitchers than non-pitchers in the average seven-game span, and through April, this year’s usage gap was the most skewed toward pitchers yet.
Granted, that’s partly a product of this year’s curtailed spring training, but the long-term trend toward reliever creep is pretty apparent. Just as the strikeout rate historically tends to rise until the league takes steps to stop it, teams historically tend to devote a growing proportion of their roster spots to pitchers—and the league hadn’t taken any steps to stop that trend until Monday. With bigger rosters in place since 2020, teams have used more than 15 different pitchers, on average, per seven-game span, and those bottomless bullpens have been better than ever. “We need an entirely new model for pitching, and it starts with not allowing teams an endless supply of pitchers,” baseball writer Joe Sheehan tweeted on Sunday. But maybe what we need isn’t an entirely new model; it’s a new way of embracing an old model. A limit of 14 pitchers could, in effect, take teams back about five years, while 13 would turn the clock back 10 years; 12 would rewind 20 years; and 11 would wipe away 30 years.
The trends toward higher strikeout counts and pitcher counts are connected, of course: The fewer innings any individual pitcher is expected to shoulder, the more effort they can expend (and the more effective they can be) on a per-batter basis. When starting pitchers were expected to pitch complete games, they had to conserve their strength. As Giants ace Christy Mathewson (or more likely, his ghostwriter) wrote in the 1912 book Pitching in a Pinch, “Some pitchers will put all that they have on each ball. This is foolish for two reasons. In the first place, it exhausts the man physically and, when the pinch comes, he has not the strength to last it out. But second and more important, it shows the batters everything that he has, which is senseless. A man should always hold something in reserve, a surprise to spring when things get tight. If a pitcher has displayed his whole assortment to the batters in the early part of the game and has used all his speed and his fastest-breaking curve, then, when the crisis comes, he ‘hasn’t anything’ to fall back on.”
Mathewson’s philosophy started to fade long ago. In David Halberstam’s 2003 book The Teammates, a St. Louis sportswriter recalls former Cardinals pitcher Harry Brecheen telling teammate Murry Dickson, who had a reputation for coasting, “Times are changing. You can’t do what you used to do any more—you’ve got to go out there and throw as hard as you can for as long as you can.” Brecheen supposedly said that in 1947. But Brecheen, at least, was still aiming to go long; today’s pitchers abide by “as hard as you can” without worrying about “as long as you can.” The “pinch” Mathewson had in mind was the ninth inning. Only once this season has a starter pitched more than eight innings.
In 2017, former Mets pitcher Ron Darling said of his pitching approach in the ’80s, “Over the course of 120 pitches in a game, I would throw max-effort fastballs maybe 10 times.” Contrast that with the recent words of 23-year-old Mariners rookie Matt Brash: “I throw my stuff full effort. I don’t take a pitch off. I don’t baby anything. I’m throwing it as hard as I can every pitch.” That approach isn’t purely a product of youth or inexperience; it’s what teams want pitchers to do. Baseball Prospectus’s Russell Carleton has likened one-inning relievers to an invasive species that has colonized the league and turned big league hitters into less imposing predators. Now, even starters are pitching a lot like relievers: With a fully-stocked bullpen behind them, it’s perfectly fine for them to go five innings, which is roughly what they’ve averaged per outing for the past few years. And if they know they don’t need to save something for the seventh (let alone the ninth), they can air it out early, a recipe for even more no-hit bids than we’ve seen in preceding seasons.
These days, the “pinch” is permanent. As former Red Sox and Cubs GM and current MLB consultant Theo Epstein told Nightengale last year, “The job evolved from trying to go into the seventh, eighth, and ninth innings to missing as many bats as you can for five innings. It’s a self-fulfilling prophecy, because now teams are completely changing their pitching development. Instead of developing starters that can go through a lineup three or four times, you’re developing pitchers who just throw as hard as possible with this crazy spin rate to miss bats.’’
Two more graphs can capture the phenomenon of pitchers putting the pedal to the metal more and more often as the pitch-tracking era (2008 to present) has progressed. The first shows the average speed difference between starters’ and relievers’ 95th-percentile (that is, fastest) four-seam fastballs and average four-seam fastballs, with individual pitchers’ influence on the leaguewide gap weighted by their number of fastballs thrown. The second shows the average speed difference between their 95th-percentile fastballs and 5th-percentile fastballs (that is, roughly their fastest and slowest, eliminating the most extreme pitches on each end of the spectrum, which sometimes indicate dodgy data). Smaller gaps between maximum and average velocity or between maximum and minimum velocity are suggestive of more max-effort pitchers and fewer pitchers who are pacing themselves to pitch in a pinch and go as long as they can.
Sure enough, both metrics show steady reductions in velocity variation, reflecting an increasing tendency to let it eat at all times. As starters have become more reliever-like, their velocity separation has shrunk even more rapidly than that of relievers. Over this 15-season span, starters have gained 2.4 mph of four-seamer velocity on average, while relievers have added “only” 1.6 mph. Pitchers aren’t throwing harder on the whole just because they’re raising the upper bounds of their velocity ranges; they’re also boosting the lower bounds by abandoning the concept of taking something off. And as pitching staffs have swollen, teams have had the luxury of giving both starters and relievers more time off between outings, which makes them even stronger when they do get into games.
All else being equal, throwing harder makes pitchers more effective and more desirable to teams, so pitchers aren’t going to throttle back on their own. Nor are teams going to tell their pitchers to take it easy on opposing hitters. This is a classic case of misaligned incentives: Pitchers are largely acting rationally—at least in the short term—by bearing down on every offering, and teams are doing the same by stocking their rosters with flamethrowers and replacing them early and often to ensure that hitters are frequently facing fresh arms. But those tactics tend to go against the audience’s best interests, which means it’s on the league to look out for the fans—and, by extension, the health of the sport. “We just want to nudge the game back into a better balance,” Epstein said last year. “The game is constantly changing, and I think for the last 10 years it’s been moving in a direction that nobody would choose on their own if they were starting from scratch. … We need to readjust the balance between batters and pitchers to create more opportunities for players to show their athleticism and for fans to get entertainment value again.”
So, how would limiting the number of pitchers on active rosters and forcing pitchers to pace themselves, go deeper in games, and have less time off between outings accomplish that? Let us count the ways.
For one thing, pitchers would have to dial down the quality of their stuff on a per-pitch basis by taking a tick or two off and throwing fewer breaking balls, which would likely amp up offense across the board. Hitters would also see the same pitchers much more often in games, allowing them to take advantage of the times-through-the-order penalty (which was just as potent in the olden days that predated max-effort pitching). Pitchers would also suffer some disadvantages that would accrue across games and series: Relievers lose their edge the more times they face the same hitters within the same series and the more pitches they’ve thrown throughout the season. And teams’ inability to hold a parade of dominant, rested relievers in the late innings of every close game would make exciting comebacks more likely.
In addition to treating fans to more balls in play, defensive chances, and baserunners, stricter pitcher restrictions would also restore starting pitchers to their traditional roles as the sport’s protagonists. As I wrote in 2018, “the starter’s starring role in the story of most games has dwindled into a supporting part in an ever-expanding ensemble.” That’s not to say that relievers can’t be riveting too, but in general, it’s more fun to follow a head-to-head duel between two starting pitchers—or between two adjusting, tiring starting pitchers and the hitters who are trying to wear them down—than early exits followed by a succession of one-inning arms attached to players most fans haven’t heard of.
The rise of the reliever has minted an army of mostly anonymous major leaguers with largely interchangeable short-burst skill sets, making it almost impossible to remember who’s on a given roster at any particular time. Reducing that contingent would allow teams to enrich baseball’s biodiversity by repopulating benches and reintroducing dedicated pinch hitters, pinch runners, defensive specialists, and platoon partners. It would also add entertainment value via a greater variety and quantity of tactical decisions. Managers (and armchair managers) would have buttons to push and levers to pull beyond signaling for the next lefty or righty out of the pen.
Plus, every decision to pull a pitcher would be freighted with implications for the rest of the game, series, and season; managers would have to consider each hook carefully lest they find themselves out of arms. Games would once again be feasts for first- and second-guessers and sources of engaging baseball debates. And if you care about this kind of thing, some increasingly unattainable milestones—300 wins, 3,000 innings, or even qualifying for the ERA title or topping 200 frames in a single season—would be back within reach. (As, perhaps, would more potential perfect games and no-hitters, at least workload-wise, though more offense early on would make no-hitter push alerts scarcer.)
Throttling pitchers would probably improve the pace of play, too, with or without a pitch clock. Time-consuming pitching changes would be considerably reduced—the number of pitcher appearances per game has roughly doubled since the start of the DH era—and pitchers might work more quickly if they weren’t trying to set a Statcast high score with every delivery. The change might trim pitch counts, too. Games have gotten longer not only because players dawdle between deliveries, but because more pitches are thrown in the average game. Relative to the late-’80s, when comprehensive pitch-by-pitch info began to be kept, nine-inning games today are about 20 minutes longer, in part because contemporary pitchers throw about 0.4 more pitches per batter faced, 1.6 more pitches per team per inning, and almost 30 more pitches per game. Batters have probably gotten more selective, but they also foul off many more pitches (because modern pitchers are so hard to hit) and spit on many more waste pitches (because strikeout-hunting hurlers are happy to stay away from the strike zone in pursuit of the thrill of the chase). With pitches at a premium in a world with aggressive roster restrictions, pitchers would feel pressure to pound the zone and pitch to contact, which would speed plate appearances along. Admittedly, a spike in offense spurred by pitcher restrictions would mean more batters faced and more pitches, but at least that added time would be more action-packed.
Finally, subtracting some bodies from the arm barn could, counterintuitively, pay dividends from a health and safety perspective, provided the roster screws aren’t tightened too quickly and teams have time to reprogram their pitchers. Managers today are dogmatic about pitch counts, and they assiduously avoid piling up pitches on young (or, for that matter, middle-aged) arms as irresponsibly as past skippers did. Research has shown that especially elevated pitch counts can jeopardize pitchers, but in shifting from a “more pitches at lower effort” approach to a “fewer pitches at max effort” approach, teams could be incurring a different kind of injury risk. Multiple studies published in the past few years suggest that while pitch velocity is only weakly correlated with elbow and shoulder torque across pitchers, it’s highly correlated with torque for any individual pitcher. As one Driveline Baseball literature review put it, “When you’re watching a pitcher throw at the high end of his velocity spectrum, you can assume that he is experiencing higher levels of torque than usual. … It’s difficult to compare torque among pitchers, but as one pitcher throws harder, he’ll experience more stress.”
In 2020, former major league pitching coach Rick Peterson said, “Every time you’re at max effort, you’re at risk of being late, meaning your arm is in improper position at foot contact.” Doing that repeatedly, he pointed out, increases a pitcher’s risk of “wearing out the tread.” Consequently, one of the studies linked above recommended that “a pitcher may be able to reduce his elbow-injury risk by deliberately varying the velocity at which he throws each pitch.” It’s difficult to compare pitcher injury rates across eras, both because good data on injuries isn’t widely available and because changes in roster composition would skew the stats, but pitchers young and old are still seemingly going under the knife as often as ever. But perhaps throwing more pitches with more varied velocity would help protect pitchers, or at least not place them in any greater danger than they already face. It might help protect hitters, too: Hit-by-pitch rates are near record highs, and while the resulting bruises and brawls may have multiple causes, max-effort pitching at the cost of command could be a contributing factor. (Brash, for one, has nasty stuff, but thus far he hasn’t controlled it.)
Baseball has changed in innumerable ways since its 19th-century codification, and mostly for the better. But it’s not for the better that rosters are overrun with relievers who’ve cannibalized their pitching progenitors. Fortunately, the fix is right in front of us: tying teams’ arms behind their backs. Lowering the now-extant pitcher limit is a safer bet and a much easier sell than, say, moving the mound back, and it’s a little less fraught than tampering with the strike zone. It would almost certainly work better (and be less intrusive) than banning infield shifts, and it’s arguably less heavy-handed, convoluted, and potentially toothless than trying to govern how many pitchers teams can use per game. In theory, even the union has reason to root for it, despite the fact that some pitchers would lose their big league gigs. The number of roster spots would stay the same, and they’d be filled by fewer of the marginal league-minimum makers whom teams treat as disposable—and whose propagation has accelerated the payroll stagnation that fueled the league’s latest labor battles. As Baseball Prospectus’s Rob Mains wrote last year, “reducing pitcher workload leads to more pitchers on rosters, who make less money than hitters anyway. It also yields less money paid to individual pitchers, since they shoulder less of the responsibility for team wins. As a result, player compensation goes down.”
Last month, an AL executive told Jayson Stark that the league hasn’t done enough to slow the reliever assembly line. “The five-option limit will play some role, and the 13-pitcher limit may impact it a little,” the exec said. “But to be honest, I don’t think this will change the way teams put pitching staffs together. As much as the commissioner’s office may try to dissuade us from doing things, there aren’t enough guard rails in place.”
But maybe the right rails are finally in place. The bumpers blocking the gutters just need to be raised—though in this case, the limits must be lowered. “Every team wants starters who can go deep,” another exec told Stark, adding, “The problem is that starting pitchers are having a harder time than ever staying healthy and carrying bulk. And I don’t know what the answer is.”
Well, whaddya know? We happen to have one right here.