Measuring Reliever Role Definition
The recent struggles of Toronto closer Jeff Hoffman and the subsequent discussions about whether it would make sense to use Louis Varland in a traditional closer role or in a more fluid high-leverage role got me interested in how strictly teams are using their bullpen in traditional roles vs. in more fluid roles. To look into this I built a metric that I’m calling Bullpen Role Index (BRI) that is intended to assess whether a reliever is used in a traditional reliever role.
The build
I used 5 traditional categories of reliever usage as per the below.
Fireman: Enters relatively early in game in high-leverage situations. Identified by entering in the 5th or 6th inning in at least 40% of appearances, with an average leverage index of 1.5 or higher.
Setup: Identified by the same leverage threshold of 1.5 or higher, but entering in the 7th or 8th inning in at least 40% of appearances.
Closer: Pure save situations. Enters in the 9th inning in at least 40% of appearances with an average LI of 1.5 or higher.
LOOGY (left-handed specialist): A left-handed pitcher deployed primarily against left-handed batters, with short outings typical of a matchup specialist. Identified by facing left-handed hitters in at least 50% of plate appearances and averaging 2.5 or fewer outs per appearance.
Long reliever: A low-leverage arm trusted to absorb multiple innings. Identified by an average leverage index below 0.75 and recording 5 or more outs in at least 50% of appearances.
40% was used as the threshold for the Fireman, Setup and Closer categories, as relievers used in these roles are typically considered to be better pitchers and it seems likely that they will pitch in a variety of circumstances. Given these cutoffs, it is possible for one reliever to be classified in two categories.
The leverage index is measured using gmLI (game leverage index at entry), sourced from FanGraphs. gmLI captures the average leverage of the game situation at the moment a pitcher entered, independent of what happened during his appearance. A gmLI of 1.5 or higher is the conventional threshold for a high-leverage situation.
Once players are categorized into a role, BRI measures how many teammates share each role. The formula is BRI = 1 / (1 + number of teammates in that role). For example, a pitcher who is the only closer on his roster scores 1.0 on that dimension. A pitcher who shares the setup role with one teammate would score 0.5 and if the setup role is shared with two teammates the score would 0.33.
Pitchers with no qualifying role receive a floor score of 0.10. They appear on the roster and contribute innings, but don't fill a defined function.
I used a cutoff of 5 appearances for the dataset.
What it can tell us
What percentage of players are being used in a defined role.
Whether pitchers are being used in high leverage without having a defined role.
Which teams have the most structured bullpens in terms of defined roles.
Which teams share certain roles between players.
Probably some other stuff.
2026 findings (as of April 25, 2026)
Some interesting findings from 2026 so far:
Teams & role structuring
The league average across all teams is that ~41% of relievers are used in a defined role as per the categories above (“role definition rate”).
Boston (75%), Arizona (71%) and Baltimore (66%) lead the league in role definition rate. Philadelphia (10%), San Francisco (12.5%) and Minnesota (14%) have the lowest role definition rate.
At this point in the season, there is no correlation between role definition rate and bullpen ERA or bullpen FIP.
At this point in the season, no team has two players filling a traditional closer role (meaning 40% of their appearances have come in the 9th inning with a gmLI of 1.5 or higher).
The Setup role is the category in which teams have the greatest number of players sharing the role, with the Athletics having 4 players used in this role and Arizona and Baltimore each having 3 players used in this role.
12 teams have only one reliever filling the role of lefty-specialist. 7 of these relievers have FIPs under 3.00 indicating a high degree of effectiveness by the relievers filling a dedicated role here.
Only 5 players have been used in a Fireman role, with almost all these players also being used in a Setup role.
10 teams have only one reliever in the Setup category, with most teams sharing this role consistently. For 8 of these 10 teams, the FIP for their sole defined Setup reliever is above 4.20.
The average leverage index at which Closers and Setup enter the game is quite similar (0.09 gmLI higher for Closers).
Player trends
Of the top 20 relievers by K%:
8 are Closers
3 are Setup
2 are LOOGYs
7 don’t have a defined role
Of the top 20 relievers by K/BB:
5 are Closers
3 are Setup
1 is a LOOGY
11 don’t have a defined role
Of the top 20 relievers by FIP:
5 are Closers
3 are Setup
2 are LOOGYs
12 don’t have a defined role.
Some standouts from players who haven’t been used in a defined role so far:
Brennan Bernardino (COL): 13 appearances, 1.16 gmLI, 2.18 SIERA, 10 K/BB, 25.6% K rate
Antonio Senzatela (COL) 8 appearances, 1.52 gmLI, 2.74 SIERA, 4.5 K/BB, 28.1% K rate
Keaton Winn (SF): 10 appearances, 0.71 gmLI, 2.10 SIERA, 6.5 K/BB, 37% K rate
Erik Sabrowski (CLE): 13 appearances, 1.30 gmLI, 2.24 SIERA, 4 K/BB, 40.8% K rate
Robert Suarez (ATL) 12 appearances, 1.23 gmLI, 2.64 SIERA, 6 K/BB, 26.7% K rate
Mason Montgomery (PIT): 10 appearances, 1.31 gmLI, 2,24 SIERA, 3.8 K/BB, 44.2% K rate
Some standouts from players 25 or younger:
Bradgley Rodriguez (SD, 22 years old, no defined role): 9 appearances, 1.09 gmLI, 2.89 SIERA, 4 K/BB, 24.5% K rate
Mason Fluharty (TOR, 24 years old, LOOGY): 13 appearances, 1.31 gmLI, 1.77 SIERA, 4.67 K/BB, 4o% K rate
Anthony Nunez (BAL, 24 years old, Setup), 11 appearances, 1.68 gmLI, 2.06 SIERA, 7.5 K/BB, 32.6% K rate
Almost every pitcher entering in high leverage situations (average gmLI of 1.5 or greater) has a defined role, with notable exceptions for Kevin Kelly (TB), Huascar Brazoban (NYM), Lucas Sims (CWS) and Antonio Sanzatela (COL), who have all appeared on average in high-leverage situations without having a defined role.
My full data (current to April 25, 2026) can be found here: https://docs.google.com/spreadsheets/d/1_QCIEynb8HnmaRqGbYqzbp_4Vw3qIGQ5I9Cz3XrXfkI/edit?usp=sharing
Will try to update it monthly over the year.