A1988
Title: Modeling race dynamics in circuit motorsport: A state-space approach for lap time decomposition
Authors: Karolina Jindrova - Prague University of Economics and Business (Czechia) [presenting]
Petra Tomanova - University of Economics, Prague (Czech Republic)
Abstract: Lap-level racing data combine precisely recorded outcomes with sources of variation that are only partly observed. A structural state-space approach is developed to decompose lap times in circuit motorsport into measured race effects and latent time-varying performance dynamics. The observation equation relates lap time to exogenous regressors representing measurable influences, while the state equation captures unobserved processes such as tire behavior, track evolution, and other persistent performance shifts. Rather than correcting lap times through separate static regressions, the approach models observed and latent effects jointly and treats decomposition as a filtering and smoothing problem. Parameters are estimated by maximum likelihood using the Kalman filter, and smoothed states provide retrospective signal extraction. Standardized One-step-ahead prediction errors are used to check serial correlation, normality, and conditional heteroskedasticity. This structure allows measured race events to be analyzed together with latent dynamics that are not directly recorded in the data. Empirical results from Formula One lap data show interpretable estimates of measured effects while preserving a flexible latent component for evolving performance dynamics.
