| Popis: |
Soccer seems simple: "you just have to score one more goal than your opponent" (Johan Cruyf). The reality is a game of complex interactions between 22 players and a ball. Teammates have to coordinate their actions in space and time while continuously adapting to the opponent and the conditions of play: something we came to know as tactics. Until recently tactics were mainly analyzed by video-analysts who manually annotated interesting moments within a game to subsequently make qualitative assessments together with the coaching staff. Due to recent technological developments and scientific advancements a whole new era of tactical analysis opened up as during most professional games position tracking data of all players, referees and the ball is automatically collected nowadays. This results in datasets of millions of datapoints per match, that with the help of state of the art techniques from computer science and data science can be used for in depth analysis of tactical patterns. Building on the aforementioned developments, this thesis presents a body of research that aims to use data and state of the art techniques to gain in-depth insights in tactics, and to develop new analytical techniques for understanding determinants of success in attacking play. The result is a new methodology to identify key contributors and team-specific blue-prints for succes, and offers a new perspective on performance analysis in professional soccer. |
