Generalizing the Elo rating system for multiplayer games and races: why endurance is better than speed

Abstract

Abstract We introduce a non-standard generalization of the Elo rating system for competitions involving two or more participants. The new system can be understood as an online estimation algorithm for the parameters of a Plackett–Luce model which can be used to make probabilistic forecasts for the results of future competitions. The system’s distinguishing feature is the way it treats competitions as sequences of elimination-type rounds that sequentially identify the worst competitors rather than sequences of selection-type rounds that identify the best. The significance of this important modelling choice is discussed and its consequences are explored. Finally, our generalized Elo system’s predictive power is demonstrated using data from Formula One racing.