FIP, fielding independent pitching, scores a pitcher on home runs, walks, hit batters and strikeouts, then adds a constant so that the league's FIP equals its ERA. We computed that constant for every season from 1970 to 2025 from SABR's Lahman Baseball Database. It ran from 2.39 in 1972 to 3.25 in 2023, a drift of 0.86. The FanGraphs Sabermetrics Library calls it "generally around 3.10". That held within 0.10 in 14 of the 16 seasons from 2010 to 2025. Within 0.05, read to three decimals, it held in 8 of those 16 and 17 of all 56.
Season constants from the Lahman data are not new. The R package baseballDBR (Kris Eberwein, last released 2017, archived on CRAN in 2023) computes them, per season or per league, with fip_values(). We used the same arithmetic to check how often FanGraphs' 2016 rule-of-thumb table fits other seasons, and to work out the league split for 2022 to 2025.
Each season's constant is league ERA minus the weighted outcomes
FanGraphs' library entry on FIP, by Piper Slowinski (last modified 8 May 2017), gives (13×HR + 3×(BB+HBP) − 2×K) ÷ IP + constant, with weights "based on the relative values of those actions with respect to run prevention". The constant is league ERA minus the same expression on league totals. Leaving balls in play out matches a finding in Voros McCracken's "Pitching and Defense: How Much Control Do Hurlers Have?" (Baseball Prospectus, 23 January 2001): a pitcher's hit rate on balls in play barely carried into the next season, while his walks and strikeouts did.
In 2025 major-league pitchers allowed 5,650 home runs, 15,379 walks, 1,928 hit batters and 19,903 earned runs and struck out 40,645 over 43,075 innings, a league ERA of 4.158. The weighted outcomes come to 1.023 per inning, so the constant is 3.135.
The totals come from the database's Teams.csv; version 2025, released 2 January 2026, carries this notice: "This database is copyright 1996-2025 by SABR, via generious donation from Sean Lahman. This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License." Our derived constants are shared under the same CC BY-SA 3.0 licence. The Lahman package for R, version 14.0-0 of 12 February 2026, describes itself as running through 2024, so 2025 has to come from the database's own files. FanGraphs' own published constants, which we did not read, may differ.
The constant ran coldest in 1972
The 1972 constant was 2.393, when league ERA, at 3.258, was the lowest of the 56. Every season from 1970 to 1993 sat below 3.00, and so did 2002. The median change from one season to the next was 0.054; the largest, in 1993, was 0.202, and the series has not dropped below 2.95 since.
Decade means warm steadily toward the 2023 high
| Seasons | Mean constant |
|---|---|
| 1970–1979 | 2.582 |
| 1980–1989 | 2.737 |
| 1990–1999 | 3.014 |
| 2000–2009 | 3.077 |
| 2010–2019 | 3.112 |
| 2020–2025 | 3.167 |
League ERA in 2019 was 4.494, above 2023's 4.326, yet the 2019 constant of 3.203 sat below 2023's 3.249. Home runs, walks, hit batters and strikeouts accounted for more of the higher ERA, and the constant takes only what they leave over.
A flat 3.10 shifts every pitcher in a season by the same amount
With a flat 3.10, every pitcher's FIP that season moves by 3.10 minus the true constant, which left it 0.707 high in 1972 and 0.149 low in 2023. The shift is uniform for all pitchers in a given season, so their rankings relative to each other don't change. What does change is how their FIP compares with ERA and with FIPs from other seasons.
Our MLB FIP calculator carries these constants for 1970 to 2025, MLB-wide or by league, and shows the fixed-3.10 shift for the chosen season. On its default line (18 home runs, 45 walks, 6 hit batters, 190 strikeouts, 180.0 innings, 2025, MLB-wide) the breakdown gives 1.300 for home runs, 0.850 for walks and hit batters and −2.111 for strikeouts; add 3.135 and FIP is 3.17, against a league average of 4.16. Switch to 1972 and the same line reads 2.43.
Before a revision on 7 November 2026, the calculator held MLB Stats API totals for 2015 to 2026. In 6 of those 12 seasons it gave a league-average pitcher two verdicts at once: average on the league line, and above or below average on FanGraphs' 2016 table. Its prose also said a given FIP "means more in a low-scoring year". The reverse holds: 3.50 is 0.99 better than the league in 2019 and 0.66 better in 2025.
Only seasons close to 2016 put a league-average pitcher on the Average row
FanGraphs' rule-of-thumb table is headed "Based on 2016 Run Environment" and runs from 3.20 for excellent through 4.20 for average to 5.00 for awful. We placed a pitcher exactly at each season's league average on the nearest row. He lands on Average in 11 of the 56 seasons, 2016 (league ERA 4.182) among them.
He reads Above average in 22 seasons, mostly before 1993, and Below average in 15, 2023 among them. In 1972 he reads Excellent. The nearest-row reading is our method; FanGraphs labels the table for 2016 and does not apply it to other years.
American League constants ran cooler in every season since 2022
The AL constant sat below the NL's in each of the four seasons from 2022 to 2025. The gap was 0.121 in 2022 and all but closed in 2023, at 0.001, before it reached 0.157 in 2024 and 0.187 in 2025. In 2025 the AL figure was 3.042 and the NL's 3.229 against an MLB-wide 3.135, so an AL pitcher reads about 0.09 higher on the MLB-wide constant than on his own league's. We attach no cause, and such gaps are not new: the leagues differed by 0.10 or more in 21 of the 56 seasons, most widely in 1970, at 0.416.
A small caveat on league splits: Teams.csv counts batters hit on a team basis. Since interleague play began in 1997, a league's HBP total is only an approximation of the batters its own pitchers hit. In 2025 AL batters were hit 998 times and NL batters 930. Misplace the whole difference of 68 and the gap moves about 0.02, on our arithmetic. baseballDBR's per-league option sums the pitching table, which avoids this.
Seasons before 1970 are left out for want of hit-batter counts
Teams.csv leaves hit batters blank for some or all teams from 1876 to 1886 and from 1911 to 1969. The term is too large to drop: it supplied 0.071 of the league's raw figure in 1970 and 0.134 in 2025.
An older form does without it. Tom Tango's undated article "Defensive Responsibility Spectrum (DRS)" on tangotiger.net writes FIP as (13×HR + 3×BB − 2×K) ÷ IP, with no hit-batter term (scarce in old records and small anyway, he says) and no additive constant; he reads the result against a league figure. He points to earlier discussion threads for the reasoning behind the weights, and the article does not say he was first.
Japan's NPB, Korea's KBO and Taiwan's CPBL would each need constants from their own league totals, which we did not open.