statistical
Pythagorean Expectation
Also known as: Pythagorean Win Percentage, Pyth
Pythagorean expectation estimates a team's true win percentage from points scored and points allowed, using an exponent calibrated per sport. It often diverges from actual win-loss record and is a stronger predictor of future performance.
Pythagorean expectation, originally developed by Bill James for baseball, predicts a team’s win rate from how many points it scores versus how many it allows. The formula generalizes across sports with sport-specific exponents.
The formula: Pythagorean win % = PF^x / (PF^x + PA^x), where PF is points (or runs) scored, PA is points allowed, and x is the empirically derived exponent. MLB uses x = 1.83. NBA uses approximately 14. NFL uses approximately 2.37. NHL uses approximately 2.05. The exponent reflects how predictive scoring differential is in that sport.
Worked example. An MLB team scores 750 runs and allows 700 across a season. Pythagorean win % = 750^1.83 / (750^1.83 + 700^1.83) = 0.534. Across 162 games, that’s 86.5 expected wins. If the team actually goes 80-82, they’ve underperformed by about 6 wins. The gap is largely luck (one-run-game variance, bullpen sequencing, hit clustering), and over a future sample they’ll likely outperform their record.
Pythagorean expectation is the foundation for fade-the-overrated and back-the-underrated team-level bets. A team that consistently outperforms its Pythagorean is either genuinely clutch (rare, hard to verify) or due for regression. Elite Picks uses Pythagorean-style win-rate baselines as one of several inputs to team-level futures and game-line projections, particularly mid-season when records have started to diverge from underlying scoring profiles and recreational bettors are still anchored to the record.
Example
An NBA team scoring 115 points per game and allowing 110 has Pythagorean win % = 115^14 / (115^14 + 110^14) = 0.668. A team going 41-41 with that scoring profile is underperforming and projects to improve.