A dynamic online nomogram to predict match outcome in the UEFA Champions League: more than meets the eye

Abstract

Abstract Background Recently, the application of interdisciplinary research methods to sports performance analysis has become a clear trend. These methods can enhance analytical techniques and provide a deeper understanding of the matching outcome.Purpose This study aimed to develop and validate a predictive model to predict match outcomes by transferring an analytical technique common to modern medicine to sports performance analysis. We would like to identify whether interdisciplinary research methods are applicable to predicting match outcomes based on historical data and what factors may affect match outcomesMethods A nomogram was generated based on lasso-logistic regression analysis to identify the potential predictors associated with match outcomes. The predictive model was built based on a nomogram, and its performance was evaluated for discrimination, calibration, and clinical utility.Results The nomogram is an effective tool for predicting match outcomes in elite soccer, owing to its higher overall performance, discrimination, and calibration of the current model. Meanwhile, the current predictive model also highlights that counterattacks, shots on target, long balls, short passes, and fouls are positively associated with match outcomes, whereas crosses and yellow cards are negatively associated with match outcomes in the UEFA champion league. A nomogram with these variables had good predictive accuracy (Brier score: 0.21, calibration slope: 1.05, c-index: 0.84)Conclusion The nomogram model showed a good predictive accuracy and discriminatory ability. The current predictive model also highlighted that counterattacks, shots on target, long balls, short passes, and fouls are positively associated with match outcomes whereas crosses and yellow cards are negatively associated with match outcomes in elite soccer. Therefore, a nomogram may be an effective tool for analyzing soccer matches. More visualization of predicting match outcome can be checked on this website (https://athletic-performance-and-data-science-lab.shinyapps.io/DynNomapp/)