How random are team sports leagues?

Competitive sports success is often attributed to a blend of luck and skill, with ongoing debates about the precise balance. This study analyzed historical rankings of major team sports leagues (three football, two basketball, and one ice hockey) to assess the extent to which competition outcomes are influenced by randomness. We created an agent-based model for sports teams, where team strength derives from the cumulative strengths of players. Disparities in team strengths influence game outcomes in the numerical simulations, mirroring Bonabeau’s concept of self-organized hierarchies in combat among animals. Like the archetype, our model exhibits a phase transition between deterministic and random phases. We calibrated the model using collected comprehensive sports data and discovered that real leagues operate near the boundary between the phases, suggesting a balanced impact of random factors and players’ abilities, with a slight advantage for the latter. • Bonabeau’s self-organized hierarchies concept is used for modeling team sports leagues. • The model exhibits a continuous phase transition as a function of the determinism parameter β . • Real football, basketball and ice hockey leagues have been found to operate in the transition area between deterministic and random phases.