Valuing passes in football using ball event data

This master thesis introduces and evaluates several models for valuing passes in football. We use event data from five seasons of matches from the top five leagues in Europe. The most simplistic model considers the value of possessing the ball in a certain area of the pitch. The other models use clustering methods to find similar passes and similar attacks to value passes. The comparison of attacks also yields the opportunity to find teams with similar playing styles. The proposed pass valuing models make it possible to rank players based on the values that were assigned to their passes. This ranking of players may help clubs to scout potential new players as well as to analyze the upcoming opponent. We show that the pass values can be used to estimate player market values and to predict match outcomes.

[1]  Jun-ichi Hasegawa,et al.  Visualization of dominant region in team games and its application to teamwork analysis , 2000, Proceedings Computer Graphics International 2000.

[2]  Jesse Davis,et al.  STARSS: A Spatio-Temporal Action Rating System for Soccer , 2017, MLSA@PKDD/ECML.

[3]  Joachim Gudmundsson,et al.  Classification of Passes in Football Matches Using Spatiotemporal Data , 2014, ACM Trans. Spatial Algorithms Syst..

[4]  Xinyu Wei,et al.  Not All Passes Are Created Equal: Objectively Measuring the Risk and Reward of Passes in Soccer from Tracking Data , 2017, KDD.

[5]  J. G. Skellam.  The frequency distribution of the difference between two Poisson variates belonging to different populations. , 1946, Journal of the Royal Statistical Society. Series A.

[6]  László Gyarmati,et al.  QPass: a Merit-based Evaluation of Soccer Passes , 2016, ArXiv.

[7]  Bjoern H. Menze,et al.  A comparison of random forest and its Gini importance with standard chemometric methods for the feature selection and classification of spectral data , 2009, BMC Bioinformatics.

[8]  Amir Rahnamai Barghi,et al.  Analyzing Dynamic Football Passing Network , 2015 .

[9]  Dino Pedreschi,et al.  The harsh rule of the goals: Data-driven performance indicators for football teams , 2015, 2015 IEEE International Conference on Data Science and Advanced Analytics (DSAA).

[10]  Luca Pappalardo,et al.  A network-based approach to evaluate the performance of football teams , 2015 .

[11]  Joachim Gudmundsson,et al.  Football analysis using spatio-temporal tools , 2012, Comput. Environ. Urban Syst..

[12]  M. Maher.  Modelling association football scores , 1982 .

[13]  Kirk Goldsberry,et al.  POINTWISE: Predicting Points and Valuing Decisions in Real Time with NBA Optical Tracking Data , 2014 .

[14]  Leo Breiman,et al.  Random Forests , 2001, Machine Learning.

[15]  Oliver Schulte,et al.  What is the Value of an Action in Ice Hockey ? Learning a Q-function for the NHL , 2015 .

[16]  Dimitrios Gunopulos,et al.  Discovering similar multidimensional trajectories , 2002, Proceedings 18th International Conference on Data Engineering.

[17]  Yisong Yue,et al.  Data-Driven Ghosting using Deep Imitation Learning , 2017 .

[18]  Xavier Anguera Miró,et al.  Automatic Extraction of the Passing Strategies of Soccer Teams , 2015, ArXiv.