Last week, the Pittsburgh Pirates demoted David Bednar, their closer, to Triple-A Indianapolis. It was an extreme and unusual move given that the team just this past winter agreed to pay Bednar, a 30-year-old veteran with 85 major league saves to his credit, $5.9 million to close games in Pittsburgh this season.
That made Bednar the fifth priciest asset on the team’s $91 million payroll ahead of such figures as Paul Skenes, Andrew McCutchen and Oneil Cruz.
Bednar earned his demotion by making three execrable appearances in Miami. In consecutive games, facing nine batters in total, he recorded just three outs, allowed four hits, walked two, allowed four runs and turned a pair of late-game ties into losses.
Coming off a 2024 season in which he ran up a 3-8 record and 5.77 ERA in 58 innings, that was enough to prompt his demotion.
The Pittsburgh Pirates, David Bednar and the "closer' fallacy
In retrospect, it is a fair question for Pirates fans to ask why the team agreed to a $5.9 million deal for a 30-year-old closer coming off such a poor 2024 in the first place. The honest answer is: that’s what teams do. They find great security in paying somebody a large sum to wear the title of ‘closer.’
The amazing part isn’t that the Pirates ponied up for Bednar, but that after having done so, they ditched him so quickly.
The mythology surrounding closers is rarely challenged, even though the David Bednars of this world routinely disprove it. This is the mythology: that there is something mystical about recording the final three outs of a game, a mysticism requiring special talents possessed by only a select few.
Two seasons ago, when Bednar was coming off back-to-back seasons of 19 and 39 saves with ERAs in the mid-twos, he was considered to be proof of this mythology. Now, Bednar having been consigned to the minors, not a single baseball expert has risen to question the legitimacy of the lore of the mystical closer.
Someone ought to, because the data disproving that myth is telling. The table below is instructive in this respect. It records average reliever performance in four categories pertinent to bullpen use during distinct eras of that use. The eras are 2016-2025, 1991-2000, 1971-1980 and 1951-1960.
The categories are save percentage, percentage of inherited runs scored, leverage index, and average number of outs recorded per reliever used. As you scan the data, ponder which era you think used bullpens most effectively. And remember, we’re looking at decade-long averages.
Seasons | Save % | Inh. Runs Sc % | Lev. index | Avg. outs |
|---|---|---|---|---|
2016-2025 | 64.0 | 32.0 | 0.987 | 3.27 |
1991-2000 | 67.6 | 32.8 | 1.020 | 3.68 |
1971-1980 | 68.5 | 34.8 | 1.037 | 5.10 |
1951-1960 | 70.5 | 36.6 | 1.030 | 5.56 |
It is beyond dispute that each of these four eras was marked by distinctive and evolving patterns of bullpen use. It is also fair to say that the data does not clearly and beyond question proclaim the modern strategy either superior or inferior to previous strategies.
But it certainly raises questions about whether all the bullpen specialization to which modern teams are committed has led to more wins, or merely to more expense and angst.
The de-evolution in bullpen strategy in high-leverage situations
Begin with the first and most basic category, percentage of saves per opportunity. For the past decade, the David Bednars of this world have saved 64 percent of their opportunities, and the percentage of saves has declined steadily as bullpen strategy has, if you will forgive the loose use of the term, ‘progressed.’
In fact, seven decades ago, when there was no such thing as "bullpen strategy", what we now call saves were successfully recorded at a rate that is six percentage points higher than today.
The modern bullpen strategy has one thing going for it: the percentage of inherited runners who score has improved, if only by a few percentage points. But a big part of the reason for that improvement is suggested in the third column, leverage index. Modern relievers, including closers, are used in less critical situations than their ancestors.
For those unfamiliar with leverage index, it measures the criticality of the moment, with 1.0 being average criticality. When we say that modern relievers are introduced at an average leverage index of 0.987, we are saying they are being eased into relatively comfortable situations, certainly relative to a quarter-century (1.020) or a half-century (1.037) ago.
If you want to know why a higher percentage of inherited runners scored in 1975 than today, that’s your answer. Modern relief strategy is designed to avoid making relievers face difficult situations.
They also pitch less. The modern reliever, on average only records a fraction more than three outs per game. That workload has steadily shrunk. And since the modern starter only averages five and one-third innings, that means more relievers – usually three or four -- are required to finish the game. In the 1970s, starters lasted into the seventh inning, leaving just 7.5 outs to be cleaned up by the bullpen, which, as the table shows, tended to cover five outs per pitcher.
In other words, it used to take just two or three pitchers to win a game. Today it takes five or six.
That in turn triggers the Bullpen Roulette principle. What’s the Bullpen Roulette principle? Simply this: the more times a manager dips into his bullpen, the more likely he is to find the guy who on that given day will blow up the game.
He might even be a $5.9 million man like David Bednar.