Large Scale Analysis of Offensive Performance in Football - Using Synchronized Positional and Event Data to Quantify Offensive Actions, Tactics, and Strategies

DSpace Repository


Dokumentart: PhDThesis
Date: 2022-02-17
Language: English
Faculty: 6 Wirtschafts- und Sozialwissenschaftliche Fakultät
Department: Sportwissenschaft
Advisor: Höner, Oliver (Prof. Dr.)
Day of Oral Examination: 2022-01-13
DDC Classifikation: 004 - Data processing and computer science
500 - Natural sciences and mathematics
510 - Mathematics
796 - Athletic and outdoor sports and games
Other Keywords:
 Sports Analytics
 Football Analytics
 Machine Learning in Sports
Offensive Performance
Football (soccer)
Tracking Data
Event Data
Order a printed copy: Print-on-Demand
Show full item record


Offensive performances in football have always been of great focus for fans and clubs alike as evidenced by the fact that nearly all Ballon d’Or winners have been forwards or midfielders. With the increase in availability of granular data, evaluating these performances on a deeper level than just goals scored or gut instinct has become possible. The domain of sports analytics has recently emerged, exploring how applying data science techniques or other statistical methods to sports data can improve decision making within sporting organizations. This thesis follows the footsteps of other sports like baseball or basketball where, at first, offensive performances were analyzed. It consists of four studies exploring various levels of offensive performance, ranging from basic actions to team-level strategy. For that, it uses a dataset part of larger research program that also explores the automatic detection of tactical patterns. This dataset mainly consists of positional and event data from eight seasons of the German Bundesliga and German Bundesliga 2 between the seasons 2013/2014 and 2020/2021. In total this amounts to 4, 896 matches, with highly accurate player and ball positions for every moment of the match and detailed logs of every action that occurred, thus making it one of the largest football datasets to be analyzed at this level of granularity. In a first step, this thesis shows how the two different data sources can be synchronized. With this synchronized data it is possible to better quantify individual basic actions like shots or passes. For both actions new metrics (Expected Goals and Expected Passes) were developed, that use the contextual information to quantify the chance quality and passing difficulty. Using this improved quantification of individual actions, the subsequent studies evaluate offensive performance on a tactical pattern level (how goals are scored) and on a strategy level (what team formations are particular effective offensively). Besides their usage on the performance side, these metrics have also been adapted from broadcasters to enhance their data story telling: Expected goals and expected passes are shown during every Bundesliga match to a worldwide audience, thus bringing the field of sports analytics to millions of fans.

This item appears in the following Collection(s)