What is Pythagorean Winning Percentage? Definition, Formula, and Example
Pythagorean Winning Percentage estimates a team's expected winning percentage from runs scored and runs allowed using the formula RS² / (RS² + RA²), revealing teams that are over- or under-performing their underlying run differential.
What is Pythagorean Winning Percentage?
Pythagorean Winning Percentage is a sabermetric formula that estimates a team's expected winning percentage based on runs scored and runs allowed. Devised by Bill James in the early 1980s and named for its resemblance to the Pythagorean theorem, Pyth W% answers a sharper question than actual W-L: how good is this team really, given the runs they put up and the runs they gave up? When a team's actual record diverges from its Pythagorean record by 4 or more wins, that team is almost always unsustainably lucky or unlucky.
How Pythagorean Winning Percentage Is Calculated
The original Bill James formula:
W% = RS² / (RS² + RA²)
Where RS is runs scored and RA is runs allowed. Multiply by games played to get expected wins.
Subsequent research showed the squared exponent slightly overstates the relationship in modern run environments. The version most commonly used today, "Pythagenpat," dynamically scales the exponent to the run environment:
Exponent = ((RS + RA) / Games)^0.287
For a typical MLB team scoring and allowing roughly 9 runs combined per game, the Pythagenpat exponent lands near 1.83. Both versions give nearly identical results in normal MLB conditions — the original 2.00 exponent is fine for back-of-the-envelope work.
A Real-World Example: The 2023 Braves
The 2023 Atlanta Braves scored 947 runs and allowed 715 over 162 games. Plugging into the basic formula:
- 947² = 896,809
- 715² = 511,225
- W% = 896,809 / (896,809 + 511,225) = 0.637
- Expected wins = 0.637 × 162 ≈ 103
Actual record: 104-58 (.642). The Braves matched their Pythagorean expectation almost exactly — they were genuinely as good as their record indicated.
Contrast that with the 2023 Arizona Diamondbacks, who went 84-78 with a Pyth record of roughly 79-83. Arizona over-performed by 5 wins, a classic signature of clutch hitting or excellent late-inning bullpen leverage. They parlayed that into a World Series appearance, but the underlying run differential said they were a roughly .500 team.
Why Pythagorean Winning Percentage Matters
Pyth W% drives evaluation across the industry:
- Regression analysis. Teams that beat their Pyth by 4+ games tend to regress the next season. The 2012 Orioles went 93-69 with a Pyth of 82-80; in 2013 they slid to 85-77.
- In-season buy/sell decisions. A .500 team with a +60 run differential is a deadline buyer; a .540 team with a −20 run differential is a deadline seller.
- Sportsbook win totals. Pre-season win totals lean heavily on Pyth-style projections rather than raw prior-year records.
- Player context. A pitcher on a team that under-performs Pyth often gets fewer wins than his ERA deserves, and vice versa.
Limitations and Common Misconceptions
Pyth W% is a useful descriptor, not a perfect predictor:
- It ignores leverage. A team that scores 10 runs in a blowout and 1 in a loss has the same RS/RA total as a team that wins three 4-3 games and loses one 9-2, but the latter wins more.
- Elite bullpens systematically beat Pyth. Late-inning lockdown arms convert close games at a rate that pure run differential under-weights. The 2014 Royals are the canonical case.
- It does not adjust for schedule strength, park factors, or injury timing. Two teams with the same Pyth record can have very different underlying talent levels.
- It is descriptive, not predictive on its own. Better forecasting models like BaseRuns and third-order winning percentage build on Pyth and outperform it for next-year projections.
Related Terms
In Legends Deck
Legends Deck team simulations use a Pythagorean-style engine under the hood: each game's outcome flows from per-inning run distributions driven by player card ratings, and season-long records track tightly to RS/RA expectations. When you build a roster in franchise mode, your projected record is your Pythagorean record — there's no hidden "clutch" multiplier. If you want to outperform your run differential, you do it the real way: stack a dominant bullpen and win the close ones.