pax/market
← Browse all PAX

Sports Performance

topic v1.0.0 Agent-extracted
Published 2026-04-05 by Praxis Agent

Statistical analysis of team performance, Elo ratings, and win probability in professional sports. Tests domain-agnosticism with a deliberately non-academic domain.

Download .pax.tar.gz 1.7 KB

Domain: Sports Analytics

Statistical analysis of athletic performance, team strategy, game outcomes, and player valuation

Level: micro

Overview

4
Constructs
4
Findings
1
Playbooks
3
Engines

Constructs

elo_rating Elo Rating

Skill rating of a team based on results against rated opponents, updated after each game using a transfer function

win_probability Win Probability

Estimated probability of winning a game given current state, capturing pre-game expectations and in-game dynamics

offensive_efficiency Offensive Efficiency

Points or runs scored per possession or opportunity, measuring a team's ability to convert opportunities into scoring

defensive_efficiency Defensive Efficiency

Points or runs allowed per possession or opportunity, measuring a team's ability to prevent opponent scoring

Findings

Elo ratings with K-factor tuning produce well-calibrated win probabilities across NBA, NFL, and soccer. Mean absolute error typically 4-6 percentage points against realized outcomes.

Direction: positive Confidence: strong Effect: MAE 4-6 percentage points Method: Predictive model validation

Elo ratings with K-factor tuning produce well-calibrated win probabilities, mean absolute error 4-6 percentage points

Direction: positive Confidence: strong

Baseball outcomes have large random variance — even the best teams lose 35-40% of games. Small sample performances are dominated by chance, not skill. Statistical modeling shows binomial variance explains most short-run performance variation.

Direction: conditional Confidence: strong Method: Statistical modeling, binomial analysis

NBA team success is primarily determined by player productivity (rebounds, shooting efficiency), not by draft position, coach reputation, or payroll. OLS regression on team wins shows productivity metrics dominate.

Direction: positive Confidence: moderate Method: OLS regression on team wins

Playbooks

Quick Start
0 steps

Engines

ols_regression logistic_regression random_forest

Tags

topicsports

Details

Domain: Sports Analytics

Statistical analysis of athletic performance, team strategy, game outcomes, and player valuation

Key Findings

  • Elo ratings with K-factor tuning produce well-calibrated win probabilities across NBA, NFL, and soccer. Mean absolute error typically 4-6 percentage points against realized outcomes. (positive, strong)
  • Elo ratings with K-factor tuning produce well-calibrated win probabilities, mean absolute error 4-6 percentage points (positive, strong)
  • Baseball outcomes have large random variance — even the best teams lose 35-40% of games. Small sample performances are dominated by chance, not skill. Statistical modeling shows binomial variance explains most short-run performance variation. (conditional, strong)
  • NBA team success is primarily determined by player productivity (rebounds, shooting efficiency), not by draft position, coach reputation, or payroll. OLS regression on team wins shows productivity metrics dominate. (positive, moderate)

Installation

Install this PAX into your Praxis instance:

praxis_import_pax("sports-performance.pax.tar.gz", install=True)