On August 15 in New York, Rays manager Kevin Cash made a momentous pitching change, pulling starter Jake Faria after 10 outs and 66 pitches. With that move, the Rays set a major league record for the most consecutive games without a single starting pitcher lasting more than five innings: 16. From July 27, when Chris Archer made his final start for the franchise, to August 18, when Archer trade return Tyler Glasnow pitched into the seventh inning, no Rays starter recorded an out in the sixth. Archer, who in late July was listed as the sole starting pitcher on the team’s official depth chart, was dealt to the Pirates at the trade deadline, which briefly left that depth chart blank. In the 18 Rays contests that followed, only two Tampa Bay pitchers who took the mound at the beginning of a game—Glasnow and Blake Snell—lasted longer than four innings.
Snell aside, the Rays’ most frequent starter this season is Ryne Stanek, who has never had an outing of more than two innings. Stanek typifies a new baseball being that tactics wonks have dubbed “the opener”: a repurposed reliever who takes the ball briefly at the beginning of the game. This allows a team to play matchups from the start and line up a second, longer-lasting pitcher to face the weak part of the opposing order when the opener’s performance is compromised from facing the same hitters for the third time in a game. The Rays-promoted idea (which may also appeal to teams because it could help suppress starter salaries) has since spread to the Dodgers and (in the minors) the Twins and, in an era when teams are increasingly concerned with avoiding the times-through-the-order penalty, seems primed to take hold elsewhere.
The average MLB starter this season has faced only 23.2 batters, and even the average excluding the outlier Rays (23.4) would be the lowest ever. Before 2014, the average number of batters faced by starters had never fallen by 0.4 or more in more than two consecutive seasons; since then, it’s plummeted by at least that much for four years in a row. As a result, starters this season have faced hitters the third time through the order or later in only 23.7 percent of their matchups, which is also an all-time low.
This year’s Rays are only the fourth team since 1908—the first year for which box scores are available via Retrosheet—to post a streak of more than 10 games without a starter pitching more than five innings. The first three teams were all terrible. The previous record holders, the last-place 2012 Rockies, tried limiting starters to 75 pitches in a futile attempt to tame Coors Field. And the last-place 1997 Athletics, who mustered a 12-game streak, got there through sheer incompetence, thanks to one of the worst rotations of the sport’s expansion era.
The last-place 1993 A’s, who became the first team to do it when they went 11 consecutive games in July without a start of more than five innings, had a lousy rotation, too. But for a span of six games during that streak, which transpired 25 summers ago, they also put in place a pioneering plan to limit their starters’ exposure by removing them early from games. In the process, they broke with baseball tradition and provided an early glimpse of future rotation tinkering. “I think it was, in retrospect … a precursor to all the things that you see today,” says TBS and SNY analyst Ron Darling, who pitched for that team. Or, as his ’93 teammate-turned-agent Bobby Witt says, “It was the tip of the iceberg.”
Longest Team Streaks Without a Starting Pitcher Going More Than Five IP
|Team Game # in Last Game of Streak
|Team Game # in Last Game of Streak
For future Hall of Fame manager Tony La Russa and his longtime pitching coach and confidant Dave Duncan, the first series after the 1993 All-Star break was the last straw. On Sunday, July 18, Oakland lost to the Yankees 13-6, ending a three-game sequence in which the visiting A’s were wiped out 32-14 after winning the first contest in the four-game series. The loss sank the team’s record to 39-49, the second-worst tally in the AL West and the third-worst in the AL overall. For the A’s, who had won three pennants in a row from 1988 to 1990 and made it to the ALCS in 1992, losing was an unfamiliar feeling; no Oakland team had finished with a sub-.500 record since 1986, and even then the A’s had gone 45-34 after La Russa took over the team in July.
Oakland’s roster in ’93 wasn’t dramatically different from the preceding season’s, but a combination of injuries (most notably to Mark McGwire), age-related declines, and free-agent departures (Dave Stewart, Mike Moore) produced dramatically different results. The team, which would finish with 68 wins, was middling on offense and below average on defense, but its primary weakness was pitching. Like the 2018 Rays, who suffered a series of rotation-depleting losses early in the year, the A’s were short on starting talent, and adversity drove innovation. “Our starting pitchers were not being aggressive enough,” La Russa recalls. “[They were] throwing way too many pitches. And so they would get to the fourth, fifth, or sixth inning and they would just tap out.” He and Duncan decided to change that when the A’s took on Cleveland on July 19.
By then, the two friends—who’d met when they played for the A’s in the 1960s—had been coaching colleagues with the White Sox and A’s for 10-plus seasons. They’d already helped refine and popularize specialized bullpens and save-centric closer usage by confining Dennis Eckersley to late-inning relief in the late 1980s, but in the rotation, they’d largely adhered to the prevailing five-man model. Neither La Russa nor Duncan recalls considering alternatives prior to 1993, although contemporary newspaper reports suggest that they’d bandied about the idea of restricting starters during idle dugout discussions in earlier years. Regardless of when the concept first surfaced, the duo had never needed to do anything drastic until ’93. But they both saw that summer as the right time to experiment, as much for motivational reasons as for the possible on-field effects. “We always lived by one of the principles of leadership: When times are a struggle, you don’t ever give into the struggle,” La Russa says. “The worst thing you can do as a leader is to say, ‘Hey, we’ve done everything we can and now it’s [up to] them and they’re not good enough.’ … There’s always something else you can do.”
In this case, “something else” was scrapping the standard five-man rotation in favor of a pitching platoon. “We got into this conversation about spring training,” Duncan recalls. “And it evolved into a conversation about, well, it seems like guys go out there for two, three innings at a time and they really have success. Or they pitch one inning at a time [and] they really do well. The conversation evolved into, ‘Why wouldn’t we do that now? We don’t really have starting pitching that’s getting us deep into the games.’”
The proposed solution to Oakland’s pitching problem, then, would look a little like a staff getting stretched out in spring training or a minor league “piggyback” or tandem rotation, which is designed to expose as many young pitchers as possible to a starting role. There was plenty of precedent for both of those things, but to bring that type of tactic to the big leagues was bold beyond imagining in 1993, a decade before Moneyball began to seed the ground for the radical thinking that led to the Rays’ recent restructuring (which Rays principal owner Stu Sternberg calls “a no-brainer”). Both La Russa and Duncan say that they must have gotten the go-ahead from A’s GM Sandy Alderson, but neither remembers a specific conversation with him about the starter-by-committee plan, and the directive came from the dugout, not from the front office. “This was back when, when you were the manager, you were the manager,” Witt says. “No, ‘We’re gonna get a lineup from upstairs and use it.’”
The experiment started following the loss on Saturday, July 17. Before the game Sunday, La Russa and Duncan informed the pitching staff that on Monday, they would assign nine of the 13 pitchers on Oakland’s active roster to three three-man platoons: Todd Van Poppel, Ron Darling, and Kelly Downs would form the first triad; Mike Mohler, Witt, and John Briscoe the second; and Bob Welch, Shawn Hillegas, and Goose Gossage the third. The remaining four pitchers—including Eckersley, the defending MVP and Cy Young Award winner—would be regular relievers. In practice, the precise deployment of pitchers would be subject to change depending on performance and game states, but in theory, no “starter” would go beyond about 50 pitches, and the three pitchers in each platoon would pitch in every third game.
“We just said, ‘Look, we’ve tried some things to get our pitching going, and this is what’s going to happen,’” La Russa says. “And, you know, you explain, but you’re not putting it up for a vote.” Van Poppel remembers the meeting much the same way. “Tony and Dunc just pulled us together and said, ‘Hey, this is what we’re doing,’” he says, adding, “We didn’t know if it was going to last one week or if it was going to last the rest of the season.” Neither did La Russa and Duncan; the A’s were in uncharted territory, and players and coaches alike were proceeding with caution, albeit also with little to lose. “The way we were performing, we would’ve tried about anything,” says Witt, who adds that doing something different from the norm can kick-start a scuffling team.
Van Poppel, who was one of baseball’s best prospects, got the scheme started on July 19 in Cleveland. He went four innings, allowed three runs, and left after 49 pitches. Darling relieved him and went the rest of the way, throwing 45 pitches. Because both players pitched efficiently and the trailing A’s didn’t need an arm for the bottom of the ninth, Downs (the third member of the first trio) didn’t get into the game. The A’s lost 4-2, but for them that was an improvement.
Van Poppel, at least, remembers the unusual arrangement as “very beneficial,” even though he had to adjust his exercise schedule so that he wasn’t working out three days in a row. “It helped me say, ‘Hey, I have to go out there and be more efficient. I gotta trust my stuff. If I want to pitch more than one inning, I better make it count,’” he says. “I liked it because I got to get back on the mound more often in game situations instead of just sitting around and thinking about it for four days like, ‘When is my next start?’ Basically, you got in, you got rested, and you got right back out there.” But his rave review wasn’t universal, in part because the A’s staff spanned more than two decades in age and experience, from the 42-year-old Gossage to the 21-year-old rookie Van Poppel. “It was probably easier for me than it was for the veterans,” says Van Poppel, who recalls that some of the more established pitchers had concerns about the system. “They were so used to the five days,” he says. “They all had routines. They were like, ‘How’s this going to play out?’”
Witt, who was 29, didn’t mind either: “It was just like throwing a bullpen, except you let it out a little bit,” he says, noting that the plan was conceived by respected, proven planners; that the staff was composed primarily of unselfish, friendly, and cooperative players; and that Welch, a former Cy Young winner, was willing to go along with it, which made it untenable for anyone else to object. But Darling, who was then an 11-year major league veteran on the verge of his 33rd birthday and hadn’t made a relief appearance since 1990, was one of the old players who weren’t thrilled by learning new tricks. “I don’t know if there was skepticism, because Tony and Dave had so much gravitas that I would’ve followed them and jumped off a bridge if they told me to,” he says. But, he adds, “Ballplayers’ minds weren’t as open as they are now.” That jibes with what a slightly fuddy-duddy-sounding Darling told reporters at the time: “I’d be remiss if I said I truly understand what’s going on.”
Darling, who describes himself as “routine-oriented,” says the new schedule was challenging to adjust to, even more so in mid-July than it would have been at the start of the season. “The relief pitchers who were used to getting up and getting in now had to have some kind of starting routine, and the starters who always had a routine to start never knew really how to get ready,” he says, recalling that he threw too many warm-up pitches before his time to come in, sapping his strength in his effort to be better prepared. “It was really my failing not to be able to make that adjustment,” he adds. “I just couldn’t do it.”
La Russa and Duncan had to adjust also, as human factors complicated the plan’s implementation. In those days, Duncan recalls, “It was all about wins and losses if you were a starting pitcher. You could have a 5 ERA, but if you had 19 wins and six losses, you were considered a good starting pitcher.” That presented a problem for the platoon plan, which La Russa acknowledges was “very unfair to the guy who starts the game.” He elaborates, “By definition, the guy who starts the game can only be the loser. He’s not gonna throw 60 pitches and pitch five innings, [so] if we’re behind he would lose, [and] if we were ahead he wouldn’t pitch the fifth inning.”
La Russa structured the plan so that the more established members of the staff—Welch, Witt, and Darling—would enter after the starter and have the best shot at a win. “The concept was maybe the first guy would give you two or three innings, then Darling would come in for the third or fourth and at least get through the fifth, maybe the sixth,” La Russa says. “The [second] guy would give you his [50-ish] pitches, and then if he had a chance to win the game you’d go to Eck.”
It usually didn’t work out that way. In the second game under the new regime, on July 20, Mohler (who more often pitched out of the pen) started and went 1 2/3, then gave way to habitual relievers Joe Boever and Kevin Campbell, who combined for the next four outs. Witt entered in the fourth and yielded two runs in four innings and 48 pitches, which might have qualified him for the win if the pitchers who preceded him hadn’t already allowed six—more than the A’s offense scored in the game. “Their intent was … just what they do today,” Darling says. “Have an opener, and then the starter would come in and pitch five or six innings, and then you would be assured of the victory. The only issue was that the team wasn’t very good, so being assured of a victory was not going to happen.”
But it did happen on the third day, July 21. That time, Downs started and threw four innings on 40 pitches, allowing only one run. Welch came in for the fifth—“It was weird coming out in a situation like that,” Downs told the press—and threw three scoreless on 45 pitches, then turned it over to Vince Horsman, Gossage, and Eckersley, who held the Indians to one run. The A’s scored seven to take it 7-2, and Welch got the win. Downs didn’t, which prompted him to say, “We are going to have to invent a new stat: NW, for nice workout.”
That was the one entry in the win column for the platoon-starter system. The next day, in Boston, Van Poppel had a decent outing as the opener, but Darling and Ed Núñez—lightly used of late, and newly nicknamed “Boom-Boom” by teammates after giving batting coach Greg Luzinski a black eye and swollen lip the night before in response to a crack about being a “once-a-week pitcher”—gave up seven runs in relief, with Darling suffering the sole blown save of his career. The A’s lost 9-7. They then dropped the Friday game to Boston, losing 6-5 in 10 innings. That one wasn’t the starters’ fault: Mohler and Witt combined for seven innings and 102 pitches, allowing only two runs, and Eckersley and Gossage got the blown save and loss, respectively. But the next time out, Downs started and allowed five runs in 4 2/3, which was all the Red Sox needed for a 5-3 win.
Six days into the new rotation order, the A’s were in last place and 1-5 since they started platooning pitchers, which signed the system’s death warrant after two turns. Sometimes the secret to survival for an unorthodox idea isn’t how well it works, but how well it works quickly. In those six games, the A’s had averaged 5.8 runs allowed. In all other games that season, they averaged 5.2 runs allowed. That was much too small a sample to draw any conclusions, and the regular relievers bore some responsibility for the runs allowed under the platoon-starter system. But the new way hadn’t wowed anyone, which it would have to do to extend its lease on life. “We didn’t get the results that we were looking for, and at the same time, we knew the pitchers didn’t like doing it,” Duncan says. On Sunday, one week after the Yankees steamrolled the A’s, Welch went 4 1/3 but threw 77 pitches, easily exceeding the platoon-period limit. The regular rotation had returned, although the A’s lost again and a frustrated La Russa punished them by removing the postgame clubhouse spread. On Monday, Darling started, went six, and got the win when the A’s scored 11 off the Angels, feasting both during and after the game.
The 1993 Oakland Athletics’ Platoon-Starter Experiment
|Longest Reliever (IP, Pit)
|Longest Reliever (IP, Pit)
|Monday, July 19
|Todd Van Poppel
|Ron Darling (4, 45)
|Tuesday, July 20
|Bobby Witt (4, 48)
|Wednesday, July 21
|Bob Welch (3, 45)
|Thursday, July 22
|Todd Van Poppel
|Ron Darling (3, 47)
|Friday, July 23
|Bobby Witt (4, 51)
|Saturday, July 24
|Joe Boever (3, 40)
La Russa says the system was flawed more in execution than in its fundamental design. Had he and Duncan designed it in the spring, it might have lasted longer, and for matchup purposes, it would have worked better with a southpaw in each triad and also in relief. (Of the 13 pitchers on the roster, only Mohler and fringe reliever Horsman threw left-handed.) Yet La Russa sees the experiment as a partial success, asserting that the starters’ strict pitch counts had helped them make “a dramatic improvement in being aggressive and going after hitters.” (The pitchers who were used as starters or long relievers during the six-game trial averaged 3.3 pitches per plate appearance over that span, compared to 3.7 for the full season.) With more effective starters, he says, “you go back to the formula that you’ve been using for 100 years—you’re not going to invent something new then.” The A’s rotation amassed a 5.43 ERA pre-platoon, with a 1.11 strikeout-to-walk ratio; post-platoon, they improved those figures to 4.85 and 1.24, respectively. Maybe that was just regression, but maybe it was a win.
When La Russa retired the platoon system in 1993, he included a caveat that it could come back. “It would be good in the right time and place,” he told the press.
Ron Kroichick of the San Francisco Chronicle, who covered the A’s for the Sacramento Bee from 1987 to 1995, says that few observers at the time thought they were getting a glimpse of where the game would be going in 25 years. “The move was viewed as a curiosity, more weird and borderline-desperate than visionary,” he says. “Nobody really thought it would stick (and it didn’t, obviously). I believe some media may have muttered about Tony trying to show off how smart he was, though in fact he and Duncan really were ahead of their time.”
Some contemporaries could see that. “La Russa’s odd-ball moves might become stuff of legend,” read a headline in the Chronicle on July 21. Below the byline, Glenn Dickey wrote, “As dramatic as the change seems now, in years to come, we may see it only as a natural progression in the way pitchers are used.” Darling describes the prescient interlude as “analytically driven,” and in a sense it was. La Russa and Duncan weren’t looking at starters’ splits each time through the order, but they were picking up on the insight those stat supply: that pitchers are better in short bursts.
In 2009, Bill James Online writer Dave Fleming laid out the statistical case for what he called the “3-3-3” rotation, making a case based on times-through splits. “The concept of the 3-3-3 Rotation, as outlandish as it may appear, is exactly where major league baseball is heading,” he wrote, not realizing that La Russa had already debuted the idea 16 years before. More recently, Russell Carleton of Baseball Prospectus has examined the platoon-starter system (and related ideas), and concluded that it could deliver a modest advantage, at least on paper.
The non-paper part is why La Russa, who bullpenned his way to a title with the Cardinals in October 2011, isn’t ready to say that the model makes sense over a full season, given a fully stocked staff. “The best way to win is to prepare [for], expect, and [encourage] the starter [to] take pride in going into the last third of the game,” he says. Relying on relievers game after game, he argues, will make them less effective, both because of batter familiarity and because of overwork. “If you try to do it because of some kind of mathematical reason that it works, then the math does not apply on a day-to-day basis.”
La Russa’s plaque in Cooperstown describes him as a “master of maneuvering lineups and managing bullpens.” One weird week in 1993 probably doesn’t deserve an amendment to that text or a sequel to Three Nights in August (Six Days in July?), but it does burnish La Russa’s “innovator” cred. His rotation machinations came too soon to set a trend, but with hindsight, they’re proving prophetic. In an era when wins aren’t king, the La Russa–Duncan brainchild looks more and more like the blueprint for progressive clubs.
The Rays weren’t inspired by the A’s experiment, but they do see some parallels between the old Oakland gambit and the approach they’re employing now. “Our focus has been on deploying our pitchers in a way that takes care of them, puts them in position to succeed, and gives us the best chance to win,” says Rays senior vice president of baseball operations Chaim Bloom. “Credit Kevin [Cash], [pitching coach] Kyle [Snyder], [bullpen coach] Stan [Boroski], and our whole staff for prioritizing the strengths of each individual pitcher, even if it meant doing things a bit differently. I’m not surprised to hear that it’s been done before—so many ‘new’ things in this game really aren’t.”
Although not every staff would benefit from being configured like these Rays or those A’s, the typical rotation is trending that way. “I would not be surprised if it [evolves] to something like that,” Duncan says. “But at that particular time, it was so different than what anybody else was doing that it wasn’t the time for it.” The time for it, it turns out, was a quarter-century later. “If this works, what a great story,” Darling said in 1993. Its next chapter is being written right now.