In this study, the Pythagorean Expectation formula, a fundamental concept in sports analytics introduced by Bill James, is applied to predict team rankings in Euroleague basketball based on total wins. The formula estimates a team's expected win percentage by analyzing runs scored and allowed, allowing for the identification of over-performing or under-performing teams. The research investigates the optimal exponent for the Pythagorean Expectation formula for each season's dataset and establishes a 95% confidence interval for this exponent using the bootstrap method. Additionally, the formula is used to predict teams' final win counts based on mid-season points scored and allowed. The study also explores machine learning methods, employing five regression models at different dataset levels. Among these, the binomial logistic regression model, using only points scored and points allowed as covariates and selected through stepwise regression based on the AIC criterion, emerges as the most accurate predictor of teams' final win counts. In summary, this study offers insights into the Pythagorean Expectation formula's application in sports analytics and its comparison with machine learning methods for ranking prediction in Euroleague basketball. While the formula provides a straightforward and reasonably accurate prediction method, more complex models outperform it, making them suitable for analysts with advanced statistical expertise, while the formula remains a viable option for those with less statistical background, offering reasonably accurate predictions.
Zoom link: https://uoc-gr.zoom.us/j/97374751400?pwd=UjZ4cUVaSXZ2bnZUT1BNSVhySFlDQT09