Why underdogs do better in hockey than basketball

About this video

• Video Title: Why underdogs do better in hockey than basketball
• Channel: Vox
• Speakers: Not specified in the transcript
• Duration: 00:07:32

Overview

This video explains the statistical reasons why underdogs might appear to perform better or why skill is a less dominant factor in hockey compared to basketball. It explores how factors like sample size, number of players, and game dynamics influence the degree to which skill or luck determines outcomes in different sports.

Key takeaways

• Skill vs. Luck Continuum: Sports can be placed on a continuum from pure skill to pure luck. The NBA is farthest from random (most skill-dependent), while hockey is closest to randomness (most luck-dependent).
• Sample Size Matters: The number of games in a season significantly impacts how much skill can emerge. Major League Baseball with 162 games allows skill to dominate, while the NFL with 16 games leans more towards luck due to a small sample size.
• Game Dynamics Influence Outcomes: Even with the same number of games (like the NBA and NHL's 82), the way the game is played affects skill's impact. Basketball's faster pace and forced shots lead to more scoring opportunities and thus a greater influence of skill, while hockey's more fluid play and less defined possessions reduce sustained control and amplify luck.
• Number of Players and Possession Distribution: Sports with fewer players (like tennis) allow individual skill to be a primary determinant. In team sports, how possessions are distributed among players is crucial. Basketball allows star players to dominate minutes, while hockey's fast pace limits even top players' ice time, diluting individual impact.
• Talent Pool and Body Types: Sports that inherently reward outlier body types, like basketball's reward for tall players, can introduce more variance. This means less skillful but exceptionally tall players can still compete, increasing the role of luck. Hockey and soccer do not require such extreme body types, leading to less variance.
• Statistical Variance: The "Pythagorean theorem of statistics" suggests that observed variance equals skill variance plus luck variance. By estimating the variance of luck (influenced by factors like the number of games), researchers can infer the variance attributed to skill.