Let’s Check In on Bullpen Usage

Last weekend’s series between the Los Angeles Dodgers and the San Diego Padres did not disappoint. It had all the energy and elements of a postseason series in April. Fernando Tatis Jr. returned from the IL to launch a home run off of Walker Buehler. Mookie Betts ended the second game in dramatic fashion with an incredible diving catch. Tempers flared in the opening game when Jorge Mateo was hit by a Dennis Santana pitch in the 10th inning. There was a little bit of everything. In so many ways, it was like a playoff series. Players and coaches on both sides held nothing back.

In the series finale, with the game tied and the Dodgers going for the sweep, something odd happened. Santana took the mound in a high-leverage situation even though Corey Knebel and Kenley Jansen were both available and had not pitched the day before. With the score tied at two, one runner aboard, and just one out, he was responsible for keeping the Padres off the board. Santana, an inexperienced middle reliever who barely made the roster, could not live up to the pressure of the moment. The Padres would go on to score three runs in the inning and win the game 5-2. The point here is not to tear down Santana. It’s that despite the intensity of a series between two teams battling for the division, sound early-season bullpen management overruled using the most proven late-inning relief options.

After the game, Dave Roberts was asked why he had Santana in the game in a high leverage situation. He responded saying, “obviously every game’s important but the health of our relievers is more important.” Injuries to pitchers are common throughout the season and the risk of injury is even greater in the first month, so Roberts’ concern is warranted and well-informed. Over the course of the three-game series between the Dodgers and Padres, a total of 19 relief pitchers were used (yes, I’m including Jake Cronenworth) accounting for 27 total appearances. Considering the caliber of starting pitching included the likes of Clayton Kershaw, Trevor Bauer, Yu Darvish, and Blake Snell, it’s a bit surprising that so much help was needed to finish off these games. (Okay, maybe not when it comes to Snell.)

The frequency of relief pitching has consistently increased over the last decade. The percent of total batters faced by relievers increased from 33.1% in the 2010 season to 42.4% in ’19. That is astounding, especially considering that throughout that span, the active rosters size limit remained capped at 25 players. To accommodate the additional workload, rosters saw some churn in their stable of relievers.

In each of the last 10 full seasons but one (2012), the total number of unique relievers used has gone up. Part of this is due to the increased rate of injury but it also reflects the increasing volume of work being done in relief. It’s one reason why active rosters expanded from 25 to 26 players beginning in the 2020 season. An additional roster spot may not seem like much, but it gives managers more flexibility with their bullpen. Dave Roberts chose to take some risk with an unproven arm like Santana rather than risk an overuse injury to experienced relievers like Blake Treinen and Corey Knebel.

To counter the flexibility afforded by the added roster spot, MLB introduced another rule designed to speed up the game by combatting the amount of relief appearances. Prior to the 2020 season, a new rule required pitchers to either face a minimum of three batters or the end of an inning prior to being replaced (exceptions are allowed for injury). The intent of the rule is to prevent specialist pitcher substitutions for the purpose of facing one batter. It would be reasonable to expect that if all else stays the same, requiring a minimum length for a pitcher’s appearance would generally increase the average length of an appearance across the population of relief pitchers. It would also decrease the overall number of appearances by eliminating some of the one- and two-batter appearances that are no longer allowed.

This hasn’t exactly played out. The average innings pitched per appearance did go up slightly in 2020 but so far this season, the rate is the same as it was in ’19. Also, while appearances aren’t on the way back down, the upward trend of the prior few seasons has plateaued as the rate of relief appearances is on par with that of the 2019 season. Those 27 relief appearances in this past weekend’s series? That’s nine per game. To put that in context with recent trends, the average game has seen 6.8 relief appearances since the 2019 season.

Relief Pitcher Usage Rates
Season Appearances/Game IP/Appearance Pitches/Appearance
2015 6.22 1.01 16.5
2016 6.30 1.04 17.4
2017 6.44 1.05 17.7
2018 6.72 1.07 17.9
2019 6.82 1.10 18.9
2020 6.86 1.12 19.4
2021 6.80 1.10 18.8

Of course, the data from the Dodgers-Padres series is unbalanced. The first game went 12 innings and each side went through nearly their entire bullpen; a total of 15 relievers pitched, including nine Padres, because Ryan Weathers was replaced after tossing 79 pitches in 3.2 innings. The remainder of the series averaged six relief appearances per game, which is much closer to the average. This demonstrates the importance of another recent rule change. After first appearing last season, the extra-innings runner on second base rule is back for 2021. A primary focus of the rule is to prevent teams from burning out their bullpen in extra innings games. The rule has faced a fair amount of criticism, but the impact of an extra innings game on a whole series was easily seen this past weekend. With more frequent use of openers and starters with lower pitch count limits such as Weathers, the impact of one extra inning game can ripple throughout an entire series.

In addition to relief pitching appearances ticking upwards, each appearance has been affected by the rate of walks and strikeouts. It’s no secret that another one of baseball’s trends is a growing propensity for the three true outcomes. The strikeout rate has increased from 18.5% in 2010 to 24.6% at this juncture in the 2021 season; here has also been a small increase in walks from 8.5% in 2010 to just over 9% so far this season. The number of pitches thrown in the majors correlates quite well with the number of walks and strikeouts.

This is a pretty intuitive relationship — walks and strikeouts by definition require a fairly high number of pitches. Nonetheless, the relationship perfectly illustrates the impact today’s brand of baseball is having on pitcher health. This season, relievers are averaging about 18.8 pitches per appearance, more than two pitches more than relievers averaged during the 2015 season (16.5), but down a bit after the rate spiked to 19.4 last year.

While relievers as a whole have accounted for a larger share of innings and total batters faced, this has not been true at the individual level. The average number of appearances in a season per reliever has actually decreased consistently since 2015. It makes some sense that to cope with the strain of an increased workload (i.e. more pitches), teams have had more pitchers shuffle in and out of the bullpen in order to meet the demands of the full season. Fewer appearances balance out the increase in pitches per relief appearance; the total number of pitches thrown per relief pitcher has remained relatively steady around 450 pitches per season.

Usage Rate per Reliever
Season Appearances/Reliever Pitches/Appearances Pitches/Reliever
2015 26.7 16.5 442
2016 25.9 17.4 451
2017 25.7 17.7 453
2018 24.9 17.9 446
2019 23.7 18.9 448

Bullpens play a much bigger role in baseball today than they did just a decade ago, with relief pitchers logging over 42% of innings so far this season. New rules put in place by MLB have had mixed effects on relief pitching. As Dave Roberts noted in his post-game Zoom meeting on Sunday, “player health” is his primary concern. Dennis Santana may not have been an option had the active roster size not been expanded to 26 players early in 2020. But since he was, it gave the Dodgers an opportunity to rest some high-leverage arms. It’s a marathon, not a sprint. By season’s end, we’ll have a much better measure of the impact of the rule changes, but we’re seeing the impact on a daily basis through decisions made in just about every game.





Chet is a contributor for FanGraphs. Prior to FanGraphs, he wrote for Purple Row. When not writing about baseball, he is a data scientist and outdoor sport enthusiast. He can be found on Twitter at @cgutwein.

newest oldest most voted
jfree
Member
jfree

This season, relievers are averaging about 18.8 pitches per appearance, more than two pitches more than relievers averaged during the 2015 season (16.5), but down a bit after the rate spiked to 19.4 last year.

And that doesn’t include the pitches thrown while warming up who never come into the game. Or the number of times during a season with warm-up-but-no-game-entry. I understand it is still possible to keep breaking relievers because you can keep pulling up more – but at some point this basically fixed 1 IP/appearance by relievers will irreparably disconnect with the reduced workload of starter