State of the art of research as well as public interest are calling for a detailed and objective scientific analysis of soccer matches. The main aim of this research is the quick and valid identification of tactical performance patterns in men`s soccer. Here, some novel objective analysis tools come into play, e.g., neural networks (for an overview, Perl & Memmert, 2012), which can identify tactical pattern based on position data. In the last couple of years, we developed a hierarchy of several artificial neural networks that allow for a rapid identification and classification of complex tactical patterns in soccer (Memmert & Perl, 2009a,b). This project was funded over four years by the German Research Foundation (DFG). Based on the position data of 22 players and the ball, we can find the characteristic movement and interaction patterns of each team and characteristic interaction patterns between both teams (Grunz et al., 2012). Characteristic means that several slightly distinct realizations of movements on the soccer field are summarized in only one movement pattern. If a team attacks always in a similar fashion, the algorithm will reduce these attacks to a pattern. For example, if a team attacks always on the left side, we obtain movement patterns describing the movements on the left wing. That means, the frequency of attacks on the wings / via the center, or the number of attacks that were conducted by means of short / long passes (always including the respective probabilities of success). Such statistics could lead to more elaborate findings than the average information that are usually discussed (e.g., percentage of ball possession) but still collected manually. Our complex characteristic patterns can be calculated automatically in a very short time (less than three seconds). In an additional step this pattern can be visualized on a drawn soccer field and be presented to coaches (Perl, et al., 2013).
© Copyright 2014 19th Annual Congress of the European College of Sport Science (ECSS), Amsterdam, 2. - 5. July 2014. Veröffentlicht von VU University Amsterdam. Alle Rechte vorbehalten.