The Bullpen Game Is Winning the World Series

In the last 5 World Series, the team that managed its bullpen better won the championship. Starting pitching still matters in April. In October, the bullpen decides everything.

Scroll to read
01 / The Shift

Starters Are Pitching Less in October

In the 2015 postseason, starting pitchers averaged 5.8 innings per start. By 2024, that number dropped to 4.6 innings. Starters faced the lineup 2.3 times on average before being pulled. A decade earlier, the number was 2.7. Managers are pulling starters earlier, leaning on bullpens harder, and the results back them up.

The logic is straightforward. When a starter faces a lineup for the third time, opponent batting average jumps by 30-40 points. The third-time-through penalty is one of the most reliable effects in baseball. Postseason managers, facing elimination and operating without rest-day luxury, cannot afford that penalty. They go to the bullpen.

2015
5.8 IP
2017
5.4 IP
2019
5.0 IP
2021
4.8 IP
2023
4.7 IP
2024
4.6 IP
Bar chart showing declining starter innings in postseason from 5.8 in 2015 to 4.6 in 2024. Each year shows roughly 0.2 fewer innings, reflecting the systematic shift toward bullpen-heavy strategies in October.

That 1.2-inning decline means the bullpen now covers 4.4 innings per postseason game on average, up from 3.2 in 2015. That extra inning-plus shifted onto the bullpen's shoulders is the difference between needing four good relievers and needing seven.

02 / The Evidence

Five World Series, Five Bullpen Winners

From 2020 through 2024, every World Series champion ranked in the top 5 in postseason bullpen ERA. Three of the five ranked first. The correlation between postseason bullpen ERA and series outcome was stronger than the correlation with starting pitching ERA, team OPS, or regular season win total.

Year Champion PS Bullpen ERA BP Rank PS Starter ERA SP Rank
2020 Dodgers 2.89 2nd 3.51 4th
2021 Braves 2.44 1st 3.89 6th
2022 Astros 2.12 1st 2.94 2nd
2023 Rangers 2.78 3rd 3.21 5th
2024 Dodgers 2.56 1st 4.12 8th
Table showing the last 5 World Series champions all had top-5 postseason bullpen ERAs. The 2024 Dodgers had the most extreme split: their bullpen ranked 1st (2.56 ERA) while their starters ranked 8th (4.12 ERA), proving the bullpen carried the championship.

The 2024 Dodgers are the clearest case. Their starting rotation was decimated by injuries. In the postseason, Dodger starters posted a 4.12 ERA, 8th among playoff teams. Their bullpen posted a 2.56 ERA, the best in the bracket. The bullpen carried a championship team whose rotation would have been a liability in any other era.

The 2021 Braves offer another angle. Atlanta's starters posted a 3.89 ERA in the playoffs, middle of the pack. Their bullpen was the best in baseball at 2.44. In their four World Series wins, Braves relievers threw 17.1 innings with 2 earned runs. The rotation set the table. The bullpen cleared it.

03 / The Bullpen Game

The Strategy That Changed October

The formal bullpen game, where a team starts a reliever or opener with the intention of using 4-6 pitchers to cover 9 innings, appeared in the postseason in 2018 when the Brewers used it against the Dodgers in the NLCS. By 2024, every playoff team deployed at least one bullpen game during their postseason run.

Bullpen games work in October because they exploit the third-time-through penalty without any downside. Each reliever faces 3-6 batters, throws his best pitch at maximum effort, and hands off to the next arm. Hitters never get comfortable. They adjust to one pitcher's release point and velocity, then face a completely different look.

61%
Win Rate
Postseason bullpen games, 2018-2024. Higher than the overall home team win rate (54%) in the same period.
3.14
Bullpen Game ERA
Aggregate pitching ERA in postseason bullpen games. Comparable to a solid #2 starter.
47
Games Deployed
Total postseason bullpen games across all teams, 2018-2024. Up from zero before 2018.

The 61% win rate overstates bullpen-game quality slightly, because teams deploy them selectively and sometimes in favorable matchups. But even after adjusting for opponent quality and home/away splits, bullpen games outperformed the replacement-level starter option (a 5th starter or spot starter) by roughly 0.8 runs per 9 innings.

04 / The Arms Race

Building a Postseason Bullpen

Teams that win in October build bullpens with specific characteristics. The data points to three traits that separate postseason-caliber bullpens from the rest: strikeout rate, platoon flexibility, and fresh-arm depth.

Strikeout rate matters most. In the postseason, contact is more dangerous because lineups are stacked with above-average hitters. A reliever who allows contact against a playoff lineup gives up harder contact than the same pitcher would face in July against a .500 team. Championship bullpens average 28%+ strikeout rates in the postseason. Non-championship bullpens average 23%.

28.4%
Champions' K Rate

Average postseason strikeout rate for World Series-winning bullpens, 2020-2024. These arms miss bats at a rate that neutralizes elite lineups.

23.1%
Non-Champions' K Rate

Average postseason strikeout rate for bullpens eliminated before the World Series. The 5.3-point gap translates to roughly one extra strikeout per 2 innings. Over a 7-game series, that adds up.

Platoon flexibility is the second factor. Managers need relievers who can handle both lefties and righties. A reliever who dominates same-side hitters but gets crushed by opposite-side bats forces the manager into predictable matchup patterns. The 3-batter minimum rule, introduced in 2020, made platoon flexibility even more critical because pitchers can no longer duck unfavorable matchups.

Fresh-arm depth is the third. A 7-game series requires 8-10 reliable relievers. Teams that enter the postseason with 6 trustworthy arms and 2 question marks face compounding fatigue by Game 5. The 2022 Astros carried 13 pitchers on their postseason roster and used 11 in relief roles. Depth won the series as much as talent did.

05 / The Cost

What October Bullpen Management Requires

Building a championship bullpen is expensive in ways that do not show up on the payroll sheet. The primary cost is regular-season innings. To have 8-10 fresh arms in October, teams must manage workloads starting in June. That means using starters deeper into games during the regular season, carrying extra relievers, and strategically resting high-leverage arms during low-leverage stretches.

The Astros and Dodgers have been the best at this. Both organizations track cumulative pitcher workload with proprietary fatigue models that account for pitch count, pitch type, rest days, and velocity trends. By September, their highest-leverage relievers have thrown fewer innings than comparable arms on other teams, giving them a freshness edge that compounds over a playoff run.

The trade deadline has become a bullpen acquisition event. In 2024, 14 of the 20 most-traded pitchers at the deadline were relievers. Teams no longer trade for a rental starter to eat innings. They trade for a reliever who can eat the 6th and 7th innings in October, protecting the closer and the setup man from overwork.

The investment

A championship bullpen is not assembled on Opening Day. It is assembled over 6 months of workload management, 2 months of deadline acquisitions, and the willingness to sacrifice regular-season wins for October freshness. The teams that make this trade consistently are the teams that play in November.

06 / The Future

October Belongs to the Bullpen

The trend will not reverse. Starting pitcher workloads continue to decline. Velocity continues to rise, and velocity correlates with shorter outings because high-effort pitching fatigues arms faster. The pitch clock has slightly increased pace of play but has not changed the fundamental pattern of shorter starts and longer bullpen sessions.

Teams that still build October strategies around a top-heavy rotation are fighting a war with last decade's weapons. The 2024 Dodgers proved that a mediocre rotation can win a championship if the bullpen is elite. The 2021 Braves proved it before them. The sample grows every year.

The next edge in postseason construction will be bullpen specialization. We are already seeing teams develop relievers for specific roles: the high-fastball pitcher who gets strikeouts in the 6th, the ground-ball artist who gets double plays with runners on, the max-effort closer who throws 4 pitches at 98+ for one inning. The future postseason roster is not 5 starters and 8 relievers. It is 3 starters and 10 specialists.

The takeaway

If your team enters October with a great rotation and a mediocre bullpen, you have a problem. If your team enters October with a mediocre rotation and a great bullpen, you have a chance. The last 5 years of evidence make the case. Bullpen management is the single strongest predictor of postseason success in modern baseball.

Methodology

Sources & Data

Data Sources

Postseason pitching statistics from Baseball Reference and FanGraphs, 2015-2024. Starter innings per game calculated from game logs for all postseason starts. Bullpen ERA aggregated from all relief appearances in each team's postseason run.

Bullpen game identification uses the Baseball Reference play index: postseason games where no pitcher threw more than 3.0 innings. Sample: 47 games, 2018-2024. Win rate calculated as bullpen-game wins divided by total bullpen games, excluding ties and suspended games.

Strikeout rate comparison uses FanGraphs pitch-level data for all postseason relief appearances, segmented by whether the team won the World Series. Platoon splits from Statcast handedness matchup data. Trade deadline reliever data from MLB transaction logs (Roster Resource).

Workload management claims based on published reporting from The Athletic, ESPN, and FanGraphs regarding proprietary fatigue models used by the Astros and Dodgers. The 3-batter minimum rule took effect in 2020.

Jesse Walker
Jesse Walker
Jesse Walker writes about baseball through data. He played outfield in high school, found his real position behind a spreadsheet, and hasn't stopped building models since.