Skip to main navigation menu Skip to main content Skip to site footer

Beyond the Scoreboard: Advancing Fairness in Athlete Selection with Simulation-Based Tournament Strategies


The process of selecting athletes for competitive sports teams is often undermined by the limitations of traditional tournament formats, which can misrepresent the true skill levels of participants. This issue is exemplified by a scenario observed in a table tennis team tryout, where a moderately skilled player advanced to the final round due to consistently facing weaker opponents, while more adept players were eliminated early against stronger competitors. Such occurrences cast doubt on the fairness and effectiveness of single elimination tournaments for player assessment. Addressing these concerns, our study conducts a thorough analysis of various tournament selection strategies, including single elimination, Swiss tournaments, and novel graph and sorting-based methods. By modeling players as Gaussian distributions with established mean skill levels, we simulate match outcomes to quantitatively evaluate the efficiency and accuracy of each strategy. Our evaluation employs two loss functions: Strict Loss, to gauge ranking precision, and Binary Loss, to assess the accuracy in identifying top performers. The experimental results reveal significant insights. Strategies integrating Elo ratings with circular graph approaches show enhanced performance, particularly in larger player groups, while TrueSkill and single elimination exhibit limitations in scalability and nuanced player ranking. The Swiss tournament, although consistent, experiences fluctuations in loss, suggesting areas for refinement. Notably, a novel graph-based strategy emerges as a stable and efficient alternative, underscoring its potential for future research. These findings aim to guide the development of more equitable and precise selection processes in sports team composition.