An Adaptive Sequence-Weighting Model for Stroke Effectiveness in Table Tennis

Table tennis is characterized by a sequential stroke structure, in which each stroke influences subsequent strokes within a rally. Traditional evaluation methods assess strokes independently and fail to capture these sequential effects. This study proposed an adaptive sequence-weighting model of stroke effectiveness (SE) by incorporating the weighted influence of subsequent strokes. A total of 100 matches involving the world’s top 25 male and female players during 2023–2024 were analyzed. Generalized estimating equations (GEEs) were used to compare SE by match outcome, gender, and association affiliation. The results show that: (1) significant differences in SE between winning and losing players were observed at all stroke positions (all adjusted p < 0.05), except for the sixth stroke among female players (adjusted p = 0.084); (2) SE declined progressively as rally length increased in both genders, with positive values during the first three strokes and negative values from the fourth stroke onward, while no significant gender differences were found at any stroke number (adjusted p > 0.05); (3) Chinese male and female players showed higher SE in prolonged rallies compared with players from other countries and regions, with significant differences observed at the fifth, sixth and after-sixth strokes (adjusted p < 0.05). These findings offer a novel framework for evaluating stroke effectiveness by accounting for sequential dependencies. Coaches can identify strengths and weaknesses at specific stroke positions and prioritize training areas based on quantified evidence. A case study using Paris Olympic match data is presented to illustrate the model’s potential practical application. Despite these promising results, several considerations should be noted: the findings are based on offensive shake-hand players and may not generalize to other playing styles, and the proposed metric remains an exploratory index that requires further independent validation.