INTRODUCTION: Para-table tennis is an opposition sport where the mind and self-confidence have a predominant role. The purpose of this study is to estimate the probability of winning a para-table tennis match and to calculate the evolution of this probability at the end of each set. METHODS: We have collected all the para table tennis matches since 2016, i.e. 26 535 mens and womens confrontations of all impairment categories. For each match, we have the exact score, the international ranking of opponents, their impairment category and the results of previous confrontations. Five ordinal logistic regression models were used to estimate the probability of final result, before the start of the match, then after each set until the end. RESULTS: Some indicators did not significantly enrich the prediction model such as the impairment category (p = 0.13), the tournament phase (p = 0.42). The main indicators retained are the international ranking of opponents (p<0.01) and the results of previous matches (p<0.01). Before the beginning of the match, based mainly on the rankings and previous confrontations, 81% of the results are correctly predicted, including 40% with the correct scores. After the first set, 83% correct predictions including 51% with correct scores. At the end of the 4th sets, the model has a success rate of 66% with great importance given to the result of the 4th set. CONCLUSION: The models used have shown very good predictive performance and are statistically robust. The study brings real added value in the prediction when the opponents have a similar level. The results of this study can be used by the staff for mental preparation before a Para-table tennis match to study the different possible scenarios.
© Copyright 2022 27th Annual Congress of the European College of Sport Science (ECSS), Sevilla, 30. Aug - 2. Sep 2022. Published by Faculty of Sport Science - Universidad Pablo de Olavide. All rights reserved.