Olympic medal prediction combining deterministic and stochastic factors

Abstract

This paper analyzes several influencing factors by constructing a predictive model for the Olympic medal table, providing valuable insights to assist national Olympic committees in optimizing their Olympic strategies. In the preprocessing stage, the study analyzed each country’s performance over the past four Olympic Games regarding gold and total medals. Countries were grouped into two categories—developed and developing sports nations — and visualized through Principal Component Analysis (PCA). This classification is a foundational element for the predictive model in the first question. The study converted Olympic medals from past Games into corresponding scores in the first question. Calculating Pearson correlation coefficients showed that the various Olympic sports can broadly be categorized into ball and non-ball. A linear regression model was then constructed to predict medals in ball and non-ball sports using the performance data from the past three Olympic Games as independent variables. In the second question, we analyzed the relationships between coaching assignments and the performance of the women’s volleyball and gymnastics teams for three pairs of countries: China-USA and Romania-USA. Through analysis, we concluded that the “coach effect” exists, with national team scores strongly correlated with outstanding coaches. Additionally, the contribution of exceptional coaches to a nation’s total medal count exceeded 0.5. In the third question, we pointed out that the prediction of Olympic medals requires careful consideration of the leader effect, host effect, candidate effect, excellent coach effect, etc., providing a reference for the Olympic Committee to designate competition rules.