Probability prediction of groundstroke stances among male professional tennis players using a tree-augmented Bayesian network

ABSTRACT

The use of different stances can provide tennis players with a tactical advantage since it enables them to cover a larger court area faster. This is especially critical since the entire stroke process takes only 1.5 seconds. However, it is unclear which stance is most suitable on the court. The purpose of the study was to predict the probability of the four stances used for forehand and two-handed backhand (2BH) in different court situations. Four influencing variables (landing zone of the ball (LZB), positioning of the player (PP), returning direction of the ball, landing zone of the returning ball) and one target variable (groundstroke stance) were collected from 3,850 successful shots at the Australian Open by a notation system to train a Bayesian network. Conditional probabilities of stance were estimated based on the two dominant influencing variables derived from Bayesian modelling. Both PP (0.53) and LZB (0.29) were identified as the most dominant influencing variables for stance selection. Probability distributions indicated that open and semi-open stances were most commonly used for forehand strokes, while closed stance was prevalent for 2BH strokes. Our preliminary findings provide insights into the court usage characteristics of the forehand and 2BH in dominant stances.

KEYWORDS:

• Notational analysis
• performance analysis
• coaching
• forehand
• backhand
• artificial intelligence

Acknowledgements

The authors wish to thank Fuzhong Li and Xie Wu for their comments on earlier drafts of the paper.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research was supported by the grant “National Natural Science Foundation of China” [11932013].