Related Experiment Videos
Understanding match performance through productivity modelling in the Chinese super league of football: a data
Liangzhu Feng1, Enze Yan2, Hongyou Liu3
1Research Center of Sports Performance Analysis, School of Physical Education & Sports Science, South China Normal University, Guangzhou, China.
Background:
Understanding match performance from a productivity perspective has become increasingly important in professional football. Although previous studies have examined technical and physical performance indicators, limited research has evaluated productivity-based efficiency across different competitive contexts in the Chinese Super League (CSL). This study aims to model and evaluate team match performance using a productivity-based analytical framework.
Methods:
Data from 1,899 matches across the 2012-2019 CSL seasons were analysed. A two-stage methodological approach was adopted, integrating canonical correlation analysis (CCA) and an input-oriented variable returns-to-scale (VRS) DEA-Malmquist model. Technical and running-performance indicators were treated as input variables, while average points per match were used as the output measure. Contextual factors, including match location and opponent strength, were incorporated. CCA revealed a strong association between the input and output variables (canonical correlation = 0.931, p < 0.001).
Results:
Technical variables-particularly ball possession and passing accuracy-were strongly associated with match performance productivity. The contribution of physical and defensive indicators varied depending on situational conditions. Higher-ranked teams tended to exhibit higher efficiency through a combination of attacking and defensive execution, whereas lower-ranked teams were more closely associated with basic technical control and physical output. Additionally, defensive actions were more relevant in matches against stronger opponents, while attacking efficiency was more prominent when facing weaker teams.
Conclusion:
Productivity-related performance patterns in the CSL appeared to be context dependent and varied according to team level, match location, and opponent strength. The findings support the application of a DEA-Malmquist productivity framework for evaluating football performance and provide practical insights for performance monitoring and tactical decision-making.
