The End of the Starting Pitcher
Since the beginning of the 20th century, starting pitchers have thrown fewer and fewer innings. But in 2017, starter usage didn’t just continue to decline; it fell off a cliff. Whether or not we realized it at the time, baseball fundamentally changed last season.
On August 18, 2012, Reds right-hander Todd Redmond allowed four runs in 3 1/3 innings and took a 9-7 loss against the Cubs in his major league debut. The result wasn’t particularly meaningful: It was Redmond’s only appearance of the season, and Cincinnati went 97-65 that year, including 12-4 overall against the tanking Cubs.
What makes that late-summer game so significant is that it was the only game that year in which someone other than Mat Latos, Johnny Cueto, Bronson Arroyo, Homer Bailey, or Mike Leake started for Cincinnati. The five-man rotation has been around in some form since the 1920s, and you’d be hard-pressed to find a better example than that Reds team. Four pitchers made at least 32 starts and threw at least 200 innings; Leake, the fifth starter, threw 179 innings in 30 starts.
Getting 161 starts and 1,015 1/3 innings out of five starters isn’t a modern record—the 2003 Seattle Mariners used only five starters, as did the 1966 Dodgers, who went with a four-man rotation, plus swingman Joe Moeller—but it’s a nearly perfect execution of the plan most teams set out to do each spring: Find five starters and throw them in order every day all season.
What separates the 2012 Reds from most of their modern contemporaries is that they defied the famous aphorism of 19th-century Prussian general Helmuth von Moltke: “No plan survives first contact with the enemy.” No rotation survives first contact with forearm tightness, comebackers, blisters, losing feel for a breaking ball, or a hundred other small perils that assail every pitcher each time through the lineup.
Getting 161 starts from five pitchers isn’t a record, but neither is Ted Williams hitting .406 in 1941 or Denny McLain winning 31 games in 1968. However, those are watershed moments, marking the final instances of statistical achievements that were once commonplace before the game evolved and they blinked out of existence. The 2012 Reds’ rotation is similarly the last of its era, as the five-man rotation—and perhaps the starting pitcher role itself—are evolving into something different.
The two World Series teams last year, the Dodgers and Astros, had 13 pitchers between them make 10 or more starts. (That doesn’t count Yu Darvish and Justin Verlander, who joined their respective teams midseason.) Four of those 13 pitchers made the All-Star team. Only one, Clayton Kershaw, qualified for the ERA title. Mike Fiers led the Astros in innings last year. Mike Fiers!
But what the Dodgers and Astros proved is that if you get a little lucky with injury timing and plan ahead, you can fill two rotation spots with three above-average pitchers without having to trawl the scarce market of above-average pitchers who can also throw 200 innings a year.
The same year that the two World Series teams ran out those extended rotations, there was a Subaru ad in which a man cleans out his old car before handing it over to his teenage daughter. Along the way, he finds odds and ends that bring to mind fond memories of her childhood. It exemplifies the idea of the family car as a thing you can rely on for years and years, then pass down to your kids. Compare that to a Formula One car, which can hit 200 miles per hour and sustain G-forces that could tear the driver’s head from his neck. Instead of lasting 15 years and needing to be serviced every 30,000 miles like a family SUV, the Formula One car needs to be serviced after every race and have its engine replaced two times a season.
The history of the pitcher is the evolution from reliable family car to Formula One racer—from the comforting reliability of the 2012 Reds to the all-out sprint of many of 2017’s best rotations.
To be eligible for the ERA title, a pitcher has to throw at least one inning for every game his team plays—162 in a normal season, fewer in case of a work stoppage or rainouts that don’t get rescheduled. In a five-man rotation, that’s a modest bar to clear: It would require 32 starts at an average of five innings each, 27 at six innings a start, and 23 starts at seven innings each. Whatever the distribution, you don’t exactly have to be Old Hoss Radbourn to pitch that much.
Putting up the qualification barrier serves a practical purpose and a secondary, perhaps more ontological purpose. The practical purpose is to weed out flukes: Nobody thinks a rookie who throws a single scoreless inning ought to show up on leaderboards. The secondary purpose is to restrict the pool of candidates to pitchers who spent all year in a rotation, not just spot starters or cameo players. One inning per team game is enough to weed out players who had two good months, but not so restrictive that it narrows the pool to only the most durable workhorses.
Or so it once was.
For most of baseball history, single-season ERA leaderboards have had somewhere between 60 and 90 names on them. As history goes on, starting pitchers pitch fewer innings per season, but beginning in 1961, the number of games on the schedule and the number of teams in the league has increased every few years. So the number of ERA title qualifiers per season had tended to be relatively stable.
There are a few troughs—in the 1940s and 1950s, pitcher usage rates were falling, but the league hadn’t expanded yet—and a few spikes in expansion years. In 2011, 93 pitchers qualified for the ERA title, tied for the sixth-highest mark since the American League was founded in 1901. In 2014, 88 pitchers qualified for the ERA title. That number dropped to 78 the next year, then 74, and in 2017, just 58 pitchers qualified for the ERA title, marking the first time in the expansion era that number dropped below 60. In other words, the last time so few pitchers qualified for the ERA title, there were only 16 teams, each playing a 154-game schedule.
The decrease in ERA title qualifiers is a multiyear trend, but last season marked a severe single-year drop-off. Somewhat unsurprisingly, innings are being spread to more and more pitchers. In each of the past five seasons, there’s been a new record for total pitchers used. Big league teams used 755 pitchers in 2017; in 2012, they used 662. In 1998, the last expansion year, 96 pitchers qualified for the ERA title, the highest total since 1915 and the third highest ever. That year, all 30 teams combined to use 557 pitchers. In 2003, when they went a whole season using just five starters, the Mariners used 18 pitchers. Last year, they used 40.
The 315 total starting pitchers in 2017 is also a league record, though that number isn’t going up as quickly as the number of total pitchers. In 2015, 313 pitchers started a game, and in 1998, 283 pitchers did. While the total number of pitchers per season has gone up 35.5 percent in the past 20 seasons, there’s been only an 11.3 percent increase in the number of starting pitchers. All those numbers point to a clear but gradual trend over the course of a decade or more, but none of them explain the precipitous drop-off in ERA qualifiers from 2016 to 2017.
In December 2016, for The Hardball Times, Jeff Zimmerman explored the increasing volume of teams using the disabled list. In 2016, players made a record number of trips to the DL and spent a record number of days on it, and Zimmerman predicted an even greater spike in 2017 after the minimum DL time was shortened from 15 days to 10 before the season. Sure enough, last season there were 659 major league DL stints, breaking the previous record by 88. The 10-day DL is particularly useful for pitchers, because it allows them to skip one start without the team having to burn a roster spot. Missing one start is a common remedy for nagging injuries, from blisters to general soreness, and with that option available, teams are less likely to ask players to play through minor injuries.
The 10-day DL is also a great way to hide a player who’s skipping a start to keep his innings total down. As much as we now take innings caps for young pitchers or pitchers returning from Tommy John surgery for granted, they’re a relatively recent invention. As recently as Stephen Strasburg in 2012 and Matt Harvey in 2015, the imposition of an innings limit on a young pitcher has led to monumental controversy. We’re just now seeing the macro-level effects of keeping developing or recently injured pitchers under 150 innings or so.
In keeping with the historic trend, starting pitchers are also on shorter leashes than ever before. In 2016, there were 985 starts in which a pitcher retired fewer than 15 batters and therefore didn’t pitch long enough to get the win. That was a record, though the other high totals had come throughout modern history rather than clustering in the 2010s. The second-highest total was in 1999, at the height of the PED era; the third highest was in 1977, when the American League added two teams, making it a seller’s market for quality starting pitching. In 2017, 1,100 starts were over in fewer than five innings—nearly a quarter of all total starters’ outings. That’s an 11.7 percent increase from the previous record, set just the year before, and a 57.6 percent increase from 2011. The fifth and sixth innings are now the responsibility of relievers whose roles didn’t exist in 2010.
From a descriptive standpoint, these changes—and the fact they’re happening so suddenly—mostly affect how the game is described statistically. The pitcher win was codified in the rule book in 1950, though wins and losses were first commonly used around the turn of the 20th century, with the statistic proposed for the first time in 1884 and tracked in 1885. In 2016, Sam Miller, then of Baseball Prospectus, suggested that MLB reduce the innings threshold for the ERA title to 130 innings, which is the equivalent of 162 in 1962, adjusted for leaguewide pitcher usage trends.
The win is all but obsolete as a measure of pitcher quality, and to a lesser extent ERA is being phased out by sophisticated estimators such as DRA and league-adjusted numbers like ERA+, which provide context the raw arithmetic of ERA doesn’t. But baseball—perhaps more than any other sport—is defined by its statistical heritage, and the game is changing so rapidly that the numbers that tell its story just aren’t keeping up. That doesn’t affect how the game is played, or even written about, though it’s not an insignificant concern. But the game is being played differently, and the change is even more dramatic than the numbers show.
Over 120 years, pitchers have gradually lost the luxury of pacing themselves, because elite hitters have evolved from Wee Willie Keeler to Aaron Judge. It’s not enough to be steady and occasionally reach back for something extra since, in 2018, just about every big leaguer can hit 20 home runs a year. That isn’t a new problem: Pitchers have had to evolve to put out more effort over fewer innings since the days of Babe Ruth. But the rate of that evolution has accelerated to the point where the bullpen phone sounds like the noise from Annihilation.
The 2012 Reds used just 17 pitchers all season—really 13, if you count only the ones who pitched 10 or more innings. They got away with that because just about everyone was healthy and effective all year, which frankly is rarer these days than winning a World Series. Of the 13 pitchers with at least 10 innings pitched, 12 had an ERA+ of 110 or better, with Leake as the only exception. Getting away with using—for all intents and purposes—13 pitchers in a season takes a tremendous amount of luck in any time period, but I’m not convinced any amount of luck could enable a team to pull that off in 2018. Some of the best teams in baseball aren’t even bothering to try anymore.
Pitchers are setting a new record for aggregate fastball velocity almost every season, and while leaguewide pitch mix numbers aren’t changing that much year-over-year, individual pitchers have chosen to throw more strenuous out pitches until they get tired. That’s how Rich Hill made his spectacular comeback, and that’s why Lance McCullers came out of the bullpen and button-mashed knuckle-curves for four innings in Game 7 of last year’s ALCS.
There are pitchers who can do that for 210 innings a year, plus another 40 in the playoffs—maybe a dozen of them in all of baseball. If there were a way to reliably develop Clayton Kershaws or Corey Klubers, someone would’ve figured it out by now. And even if by some monumental feat of scouting, player development, and luck, you manage to get four or five of those guys on one team at the same time, there’s no guarantee they’ll continue to perform at that level for very long. Remember how set the Mets’ rotation looked after the 2015 World Series? How many 200-inning seasons do you think Steven Matz and Matt Harvey have in front of them?
That’s not to say it’s impossible to create a very good traditional five-man rotation, but it takes some doing. The Diamondbacks had to throw $200 million at Zack Greinke, trade for Taijuan Walker right before his bounce-back season, and teach Robbie Ray how to throw strikes. The Nationals threw the bank at Max Scherzer, spent the no. 1 overall pick on the best college pitching prospect ever, traded four prospects for Gio González, and still wound up starting Edwin Jackson down the stretch. It’s not easy.
So while you can’t exactly go out and get an ace (the Astros’ theft of Justin Verlander notwithstanding), you can build ace-level production out of multiple pitchers who either aren’t durable enough to make 30 starts or versatile enough to get through a lineup more than twice. The 2015 Royals got by with an iffy starting rotation by having a deep bullpen that could take the middle innings from below-average starters like Jeremy Guthrie. At the time, that didn’t look like it would work across the rigors of the regular season, since it would put too much strain on the bullpen, but once teams started trying it, it turns out you can get away with five-inning starters in a couple of places in the rotation, particularly now that bullpens go eight or even nine pitchers deep. Consider Cleveland’s Mike Clevinger, who had a comparable ERA- to Chris Sale last year, but did it while averaging fewer than six innings a start.
McCullers, meanwhile, is an archetype—a small, high-effort righty with no. 1 starter stuff. When McCullers is on, he’s outstanding: a 3.05 ERA and 10.4 K/9 ratio before the All-Star break last year, and a 2.67 career playoff ERA. The problem is he isn’t always on, which is why he finished last season with a 93 ERA+ and found himself outside the Astros’ initial playoff rotation.
Amateur pitchers like McCullers pretty much grow on trees, and once in a generation a Pedro Martínez or Tim Lincecum will show that he can maintain that ace level for 30 starts year after year, but most pitchers like McCullers either move to the bullpen, get hurt, or wash out. As recently as a year or two ago, McCullers looked like a very good future closer, and that’s exactly where he’d be going if he had to throw 200 innings a year in order to start. But that’s not the case anymore. You can’t go out and get Martínez, but you can go out and get a McCullers—or even two or three of him. In fact, the Astros spent their first-round pick last year on a similar pitcher, North Carolina right-hander J.B. Bukauskas.
Five years ago, the goal was to get 1,000 innings from five starting pitchers. Now, with the 10-day DL and 13-man pitching staffs, you can hand off 100 of those innings to the bullpen and split the remaining 900 among seven or eight starters. Not only are pitchers throwing harder and striking out more hitters than ever before, it’s a pragmatic workaround to the fact that pitchers get hurt, because teams will always have a spare.
Given the rapid rate of change in pitcher usage just over the past few years, and the unprecedented rate at which the people who run baseball teams are learning about baseball, it’s difficult to predict how the game will evolve to accommodate the new model of a starting rotation. Perhaps the de facto eight-man rotation of McCullers types is the way of the future. Perhaps we’re headed for a college-style pitching staff, with two or three traditional seven-inning starters, backed up by a bevy of swingmen who can go once or twice through the order. Or maybe—though this is unlikely, given the deist tendencies of the people who run baseball—some rule change will swing the balance back toward contact hitters, reverse a century of rising strikeouts, and allow pitchers to pitch more innings.
Either way, the current pattern of starting pitcher usage, even if it is only a stepping stone to something else, is a radical departure from what we’ve understood to be a starting rotation for the past 120 years. And our language and expectations—like opposing hitters—are yet to adapt.