Analysis of Combined Events Scoring systems : Models for comparing Athletic performances

Details

Supervisors

Hafner, Christian

Faculty

Faculté des sciences

Degree label

Master [120] en science des données, orientation statistique, à finalité spécialisée

Abstract

enThis master thesis reviews and improves the decathlon scoring system using ten years of French track-and-field data (2015-2025) to better fit performance distributions than previous World Athletics formulas. Multiple estimate techniques, including maximum likelihood estimation, quantile matching estimation, and Bayesian inferences, are used to generate six candidate distributions, Weibull, Log-Normal, Gamma, Generalized Gamma, Burr XII, and Inverse Burr. All models are then statistically evaluated using classical parametric tests and derived non-parametric test. A new scoring approach is then introduced, mapping each performance to points via the complementary cumulative distribution function (CCDF) of the chosen distribution, ensuring that points reflect the true likelihood of the performance. Applied to Olympic decathlon datasets, this CCDF-based system aim to deliver more equitable, data-driven scores compared to existing tables.