Modelling swimmers' speeds over the course of a race

A stochastic model of swimming speed over the course of a male 200m freestyle swimming race is proposed. It builds on a dynamical model reflecting the trade-off between drag and propulsion in swimming. The parameters of the model are estimated from elapsed time data observed at several points along the pool. The model fits the data well and also provides a good description of the swimming strategies of each swimmer from phase to phase in the race. An individual factor measuring how much faster or slower the individual swims relative to the average swimming speed of the race is simultaneously estimated in the course of fitting the model. This factor is, as expected, closely related to the final outcome of the race.