Team defense evaluation in handball based on event prediction with machine learning
DOI:
https://doi.org/10.51094/jxiv.1144Keywords:
machine learning, sports, artificial intelligence, AIAbstract
Handball is a globally popular team sport where two teams of seven players each compete to score goals using their hands. While recent advances in sports analytics have led to the development of quantitative performance evaluation methods across various sports, defensive evaluation methods in handball remain insufficiently established. This study proposes H-VDEP (Valuing Defense by Estimating Probabilities in Handball), a novel metric for quantitative evaluation of defensive performance in handball that utilizes tracking and event data to assess the diversity and complexity of defensive actions that traditional simple metrics fail to capture. This methodology adapts the soccer VDEP framework to handball's specific characteristics, integrating key elements such as fouls, fast breaks, and conceded goals to enable more comprehensive defensive evaluation. We developed prediction models for various defensive aspects using 81 features, including player positions, velocities, and ball-related metrics, derived from tracking data of five German Bundesliga matches. We employed CatBoost as the prediction model and optimized input features through recursive feature selection using Optuna. Performance evaluation using German Bundesliga match data achieved F1 scores of $0.379 \pm 0.145$ for goal prediction, $0.159 \pm 0.151$ for foul prediction, and $0.153 \pm 0.146$ for fast break prediction. The effectiveness of H-VDEP was verified through both quantitative and qualitative evaluations. Quantitative evaluation involved comparative analysis with actual match outcomes, while qualitative evaluation included detailed analysis of individual defensive plays. The individual play analysis examined successful and unsuccessful defensive cases, investigating their relationship with H-VDEP scores. Specifically, we analyzed defensive successes and failures across various score ranges to verify the model's ability to appropriately evaluate defensive situations. All of the code used in the experiments is available on the following page (https://github.com/sflren6741/h-vdep). This research expands the possibilities of handball analysis by providing a quantitative framework for defensive evaluation, potentially influencing tactical decision-making and player development strategies.
Conflicts of Interest Disclosure
There is no conflict of interest.Downloads *Displays the aggregated results up to the previous day.
References
R. Pollard and C. Reep, “Measuring the effectiveness of playing strategies at soccer,” Journal of the Royal Statistical Society. Series D (The Statistician), vol. 46, no. 4, pp. 541–550, 1997.
R. Pollard, J. Ensum, and S. Taylor, “Estimating the probability of a shot re- sulting in a goal: The effects of distance, angle and space,” Int. J. Soccer Sci., vol. 2, 2004.
J. Fern´andez, L. Bornn, and D. Cervone, “Decomposing the immeasurable sport: A deep learning expected possession value framework for soccer,” in 13th MIT Sloan Sports Analytics Conference, vol. 2, 2019.
D. Cervone, A. D’Amour, L. Bornn, and K. Goldsberry, “A multiresolution stochastic process model for predicting basketball possession outcomes,” Journal of the American Statistical Association, vol. 111, no. 514, pp. 585–599, 2016.
O. M¨uller, M. Caron, M. D¨oring, T. Heuwinkel, and J. Baumeister, “Pivot: A parsimonious end-to-end learning framework for valuing player actions in hand- ball using tracking data,” in International Workshop on Machine Learning and Data Mining for Sports Analytics, 2021, pp. 116–128.
M. Adams, A. David, M. Hesse, and U. R¨uckert, “Expected goals prediction in professional handball using synchronized event and positional data,” in Pro-ceedings of the 6th International Workshop on Multimedia Content Analysis in Sports, 2023, pp. 83–91.
A. Mortelier, F. Rioult, and J. Komar, “Design of a handball tactics observatory based on dynamic sub-graphs,” in International Sports Analytics Conference and Exhibition, 2024, pp. 149–166.
K. Toda, M. Teranishi, K. Kushiro, and K. Fujii, “Evaluation of soccer team defense based on prediction models of ball recovery and being attacked: A pilot study,” Plos one, vol. 17, no. 1, e0263051, 2022.
H. Biermann, J. Theiner, M. Bassek, D. Raabe, D. Memmert, and R. Ewerth, “A unified taxonomy and multimodal dataset for events in invasion games,” in Proceedings of the 4th International Workshop on Multimedia Content Analysis in Sports, 2021, pp. 1–10.
I. Mchale and P. Scarf, “Modelling soccer matches using bivariate discrete dis- tributions with general dependence structure,” Statistica Neerlandica, vol. 61, pp. 432–445, 2007.
I. Mchale, P. Scarf, and D. Folker, “On the development of a soccer player per- formance rating system for the english premier league,” vol. 42, 2012, pp. 339– 351.
L. Pappalardo, P. Cintia, P. Ferragina, E. Massucco, D. Pedreschi, and F. Gi- annotti, “Playerank: Data-driven performance evaluation and player ranking in soccer via a machine learning approach,” ACM Trans. Intell. Syst. Technol., vol. 10, no. 5, 2019.
T. Decroos and J. Davis, “Player vectors: Characterizing soccer players’playing style from match event streams,” in 2020, pp. 569–584.
D. Cervone, A. D’amour, L. Bornn, and K. Goldsberry, “Pointwise: Predicting points and valuing decisions in real time with nba optical tracking data,” vol. 28, no. 3, 2014.
Y.-H. Chang, R. T. Maheswaran, S. J. J. Kwok, T. Levy, A. D. Wexler, and K. Squire, “Quantifying shot quality in the nba,” 2014.
P. Lucey, A. Bialkowski, M. Monfort, P. Carr, and I. Matthews, “”quality vs quantity”: Improved shot prediction in soccer using strategic features from spa- tiotemporal data,” 2015.
K. Fujii, Y. Inaba, and Y. Kawahara, “Koopman spectral kernels for comparing complex dynamics: Application to multiagent sport plays,” in ECML/PKDD, 2017.
K. Fujii, Machine Learning in Sports: Open Approach for Next Play Analytics. 2025.
T. Decroos, L. Bransen, J. Van Haaren, and J. Davis, “Actions speak louder than goals: Valuing player actions in soccer,” in Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining, 2019, pp. 1851–1861.
I. Simpson, R. J. Beal, D. Locke, and T. J. Norman, “Seq2event: Learning the language of soccer using transformer-based match event prediction,” in Proceed- ings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2022, pp. 3898–3908.
C. C. Yeung, T. Sit, and K. Fujii, “Transformer-based neural marked spatio temporal point process model for football match events analysis,” arXiv preprint arXiv:2302.09276, 2023
Y. Luo, O. Schulte, and P. Poupart, “Inverse reinforcement learning for team sports: Valuing actions and players,” in Proceedings of the Twenty-Ninth Inter- national Joint Conference on Artificial Intelligence, IJCAI-20, Main track, 2020, pp. 3356–3363.
P. Rahimian and L. Toka, “Inferring the strategy of offensive and defensive play in soccer with inverse reinforcement learning,” 2021. [24] G. Liu and O. Schulte, “Deep reinforcement learning in ice hockey for context- aware player evaluation,” 2018, pp. 3442–3448.
G. Liu, Y. Luo, O. Schulte, and T. Kharrat, “Deep soccer analytics: Learning an action-value function for evaluating soccer players,” Data Min. Knowl. Discov., vol. 34, no. 5, pp. 1531–1559, 2020.
N. Ding, K. Takeda, and K. Fujii, “Deep reinforcement learning in a racket sport for player evaluation with technical and tactical contexts,” IEEE Access, vol. 10, pp. 54 764–54 772, 2022.
H. Nakahara, K. Tsutsui, K. Takeda, and K. Fujii, “Action valuation of on- and off-ball soccer players based on multi-agent deep reinforcement learning,” IEEE Access, vol. 11, pp. 131 237–131 244, 2023.
Downloads
Posted
Submitted: 2025-03-25 07:34:29 UTC
Published: 2025-03-27 09:31:38 UTC
License
Copyright (c) 2025
Ren Kobayashi
Keisuke Fujii
This work is licensed under a Creative Commons Attribution 4.0 International License.
