Predication of Final Medal Counts in Olympic Games by Monte Carlo Simulations

TL;DR

The paper tackles predicting final Olympic medal counts by integrating historical performance with program-strength signals across events. It introduces a per-event strength score $P_{ix}=\sum^n_{j=1}\frac{0.8}{2^j}(a G_{ijx}+b S_{ijx}+c B_{ijx}+d Y_{ijx})$ and a top-score based medal allocation $M_{ix}$, yielding $PM_x=\sum_i M_{ix}$, with $E=\sum_x (PM_x-AM_x)^2$ minimized to learn $a,b,c,d$. The authors validate the approach using the Paris 2024 final medals $AM_x$ and perform Monte Carlo searches over $a,b,c,d\in[0,10]$ to identify favorable configurations, reporting two data-window results (1980–2020 and 2000–2020). Predicted medal tables for 2028 are produced using the best constants, with explicit national rankings and notes on data limitations (e.g., exclusion of the host's five new sports), illustrating a practical forecasting tool for Olympic planning.

Abstract

In the paper, a program strength model was proposed to evaluate the performance of countries across different Olympic events. The model assessed how strong a country's program was in each event and also factored in the influence of past Olympic performances. The final medal counts from the Paris 2024 Olympic Games were used to validate the model and to determine the optimal set of constants using Monte Carlo simulation. Based on this model, a prediction of the final medal counts for the 2028 Olympic Games is also provides for reference.