March 20, 2017
Has the Modern Bullpen Killed Late-Inning Comebacks? (Part One)
Note: This article (and the one that will follow it later this week) is based on a presentation I made at the Society for American Baseball Research Analytics Conference in Phoenix. You can hear the audio here, and follow along with the slides here, should you desire. Fortunately, there is no video.
Part One: The Modern Bullpen
The modern bullpen is generally traced to 1988. That year the Athletics, under manager Tony La Russa, won 104 games en route to the American League pennant. Dennis Eckersley led the American League in saves with 45 and finished second in the Cy Young voting and fifth for MVP. The year before, Cincinnati’s John Franco had set a record by recording 24 saves in games in which he pitched exactly one inning. (Surprisingly, we link the modern bullpen with the American League-winning team with a Hall of Fame reliever managed by a Hall of Famer, rather than an 84-78 Reds team managed by Pete Rose.)
Prior to Franco in 1987, only five pitchers had totaled as many as 15 one-inning saves in a season: 15 by Clay Carroll in 1972; 16 by Fred Gladding in 1969, Wayne Granger in 1970, and Lee Smith in 1986; and 17 by Dave Smith in 1986.
Eckersley had joined the A’s a year earlier, after 12 seasons as a starting pitcher averaging 30 starts per year. He was a swingman in 1987, starting two games and entering as a reliever in the fourth inning (nine times), fifth inning (twice), sixth inning (seven times), seventh inning (seven times), eighth inning (12 times), ninth inning (12 times), and extra innings (four times).
He became the A's primary closer in mid-June, earning 15 of the team’s 24 saves for the remainder of the season. Only four of his saves were of the one-inning variety in 1987, and two of those were in extra innings, so only twice did he enter a game to start the ninth inning and earn a save. But La Russa must’ve liked what he saw, as Eckersley became a one-inning closer:
Though Eckersley led the majors in one-inning saves only once, in the pivotal 1988 season, La Russa and he ignited a trend:
One-inning, ninth-inning closers led to one-inning, eighth- and seventh-inning specialists as well. In 1988, there were 1,903 one-inning relief appearances out of 7,331 in total, or 26 percent. In 2016, there were 7,869 one-inning relief appearances out of 15,307 in total, or 51 percent, nearly double the 1988 rate. One-inning relievers have become the norm.
More relievers pitching shorter stints has yielded better bullpen performance, as the table below illustrates.
The difference in ERA and strikeout rate between starters and relievers has widened by about 50 percent since the pivotal 1988 season.
Comebacks in the late innings provide some of baseball’s most memorable moments. From Bobby Thomson’s Shot Heard ‘Round the World to Kirk Gibson’s pinch-hit home run (Off Eckersley! In a one-inning save attempt!) in the 1988 World Series to The Double by Edgar Martinez, a team coming back in late innings is a special thrill. With a growing proportion of innings handed over to an increasingly proficient cadre of relievers, the comeback would appear to be imperiled.
This concern became acute after the 2015 season. That year, only 11.8 percent of teams trailing after six innings came back to win the game, as did 7.4 percent of teams trailing after eight innings and 3.5 percent trailing after nine. All three figures represent post-World War II lows. The season concluded with the World Series victory of the Royals, a team whose bullpen seemingly shortened games to six innings, with Kelvin Herrera in the seventh, Wade Davis in the eighth, and Greg Holland in the ninth (until Holland was injured, when the buzzsaw became Ryan Madson, Herrera, and Davis).
As a result, baseball was seen as a game in which teams had to score early, because the modern bullpen was shutting down late-inning scoring. This perception was borne out by the numbers, with late-inning comebacks apparently in a steady (if slow) decline:
On April 5, 2016, Tom Verducci of Sports Illustrated, taking note of these trends, wrote:
Two days later, our own Russell A. Carleton responded. He noted that if one selects only games in which a team is leading by three or fewer runs after the seventh or eighth innings—i.e., games in which one could reasonably hope for a comeback—the percentage of comebacks was indeed down in 2015 but still more or less within a 40-year range:
Further, Carleton noted that with overall scoring down, there are more opportunities for teams to come back late in games, since leads are smaller. He illustrated this with a graph showing the frequency with which games entered the ninth inning with teams separated by three runs or fewer:
Going into the 2016 season, Carleton's analysis indicated that the shape of comebacks has changed this decade. Less scoring has yielded more comeback opportunities, while strong bullpens have suppressed comeback success. Multiplying more chances by less success, though, had yielded a stable number of comebacks. The difficulty of mounting a comeback against the modern bullpen of pitchers who throw 15-20 of their best pitches (the average reliever threw 17.4 last year) for one inning and then leave the game would appear to be the La Russa/Eckersley legacy.
Still, 2015 did represent a low-water mark. In 2016, a spike in home runs resulted in a 0.23 increase in runs scored per game, the largest increase in a decade. I’ll wrap up this analysis later this week by looking at comebacks in 2016 and what else has changed, and hasn’t changed, since Tony La Russa started using Dennis Eckersley exclusively to get the 25th, 26th, and 27th outs of the game.