Pythagorean Expectation in Basketball: How Points Reveal True Team Strength

Published 9 August 2026Features

A basketball team's expected winning percentage is calculated through the formula PF^13.91 / (PF^13.91 + PA^13.91), applying an exponent far larger than the power of 2 used in baseball. By taking total points scored and points allowed across a schedule, the calculation bypasses win-loss records to estimate how many games a team should have won based purely on scoring margin.

The resulting figure frequently sits beside actual records in modern league tables. When the two numbers diverge, the gap provides a diagnostic measure of a roster's underlying performance.

Calculating Expected Wins From Points Scored and Allowed

The mathematical architecture of the basketball model relies on two team statistics: points for (PF) and points against (PA). The system expresses expected win percentage through the equation:

W% = PF^k / (PF^k + PA^k)

In this formula, k represents the sport-specific exponent, set to 13.91 for the National Basketball Association. The resulting percentage is strictly bounded between 0 and 1, producing a nonlinear curve that maps scoring margins directly to expected outcomes.

To turn this percentage into expected wins, analysts multiply the expected win percentage by the total number of games played. Expected Winning Percentage (EWP) is defined as the percentage of games a team should win given its scoring balance, with the exponent within the equation denoted as β.

Because the relationship is nonlinear, adding ten points to an offensive total does not produce a flat increase in expected wins across all scoring tiers. The shift in expected win rate depends on the baseline ratio between points scored and points allowed. A team scoring 112 points per game while allowing 108 will produce a distinct expected win output compared to a team scoring 102 and allowing 98, even though both carry an identical plus-four scoring differential.

Why Basketball Demands a High Exponent

The basketball exponent of 13.91 stands apart from every other major team sport. Football uses an exponent of approximately 2.37, while hockey operates around 2.05. Baseball, where the concept originated under researcher Bill James, relies on an exponent of 2.

`` Sport Pythagorean Exponent Basketball 13.91 Football 2.37 Hockey 2.05 Baseball 2.00 ``

The exponent governs how sharply scoring differential converts into projected victories. In sports with low scoring frequency, such as hockey or soccer, a margin of one goal represents a vast share of the total game scoring. In those environments, a small numerical exponent is sufficient to separate winning teams from losing teams.

Basketball presents the opposite scoring environment. Teams routinely score more than 100 points per game, meaning a single basket constitutes less than two percent of a team's total offensive production. If the formula applied baseball's exponent of 2 to NBA scoring totals, a team outscoring opponents 110 to 100 would generate an expected win percentage barely above.545.

Raising the exponent to 13.91 steepens the curve. Under the higher power, that same scoring balance translates into a dominant projected record. The specific NBA value of 13.91 is attributed to empirical fitting against historical league scoring data to match actual win distributions.

Measuring Overperformance and Underperformance Through Residuals

Once expected wins are calculated, comparing that total to actual wins yields what MetricGate designates as the residual:

Residual = Actual Wins - Expected Wins

This arithmetic difference tracks the degree to which a team overperformed or underperformed its scoring profile over a completed schedule.

A positive residual indicates that a team accumulated more victories than its aggregate point differential suggested. A negative residual indicates that a team dropped games despite outscoring opponents across the aggregate schedule.

If a team wins 48 games in an 82-game season while its scoring differential produces an expected total of 42 wins, the positive residual of six wins highlights a team that clustered its points efficiently into close wins while absorbing heavy losses in its defeats. Conversely, a team that builds blowout wins but loses close finishes will generate a negative residual, accumulating fewer actual wins than its points imply.

The residual serves as a descriptive metric of efficiency rather than an absolute judgment on team quality. It reports how actual standings positions deviate from point distributions, establishing whether a win total was built on consistent multi-possession dominance or razor-thin margins.

The Sabermetric Foundation of the Formula

The foundation of the basketball formula comes from baseball sabermetrics. Bill James introduced the original calculation to assess baseball standings.

The baseball formula used runs scored (RS) and runs allowed (RA):

EXP(W%) = RS² / (RS² + RA²)

James termed the equation "Pythagorean" due to the visual resemblance of the squared denominators to the Pythagorean theorem of geometry, though the formula shares no conceptual connection to right-angled triangles.

Adapting James's baseline to basketball required adjusting for pace and scoring volume while maintaining the core mathematical property: an S-shaped curve that approaches zero as points allowed grow infinitely large and approaches one as points scored outpace defense. By substituting points for runs and scaling the exponent from 2 to 13.91, the adaptation retains James's core insight that points scored and surrendered provide a more stable description of team capability than game-by-game results.

The open technical question for analysts remains whether a single fixed exponent of 13.91 remains optimal as league-wide scoring averages and three-point attempt rates shift over time.