Two Universal, Probabilistic Measures of Competitive Imbalance
27 Pages Posted: 5 Oct 2014 Last revised: 10 Dec 2015
Date Written: October 3, 2014
I propose two universal measurements of competitive imbalance. The first is, like those in the existing literature, based on team season win-loss records. The second is based solely on the outcomes of the various series of individual match- ups during a season. Both measures will on average equal the expected value of (2p-1)^2, where p is the “typical” probability that the stronger team will win any particular game. Both measures will average zero when a league is competitively balanced, and will converge to 1 as the league approaches perfect imbalance. And both measures are universal, in that they are on average independent of the number of teams in the league and the number of games played during the season.
Keywords: Sports, Sampling
JEL Classification: C46, L83
Suggested Citation: Suggested Citation