AUTOMATIC, COMPLEX ANALYSIS OF TOP EU- ROPEAN SOCCER GAMES: RELATION TO LAB PERFORMANCE DIAGNOSTICS

For many years there are well developed research tools and many results with respect to the testing of human per-formance in the lab, also for soccer players (Bangsbo et al., 2006; Krustrup et al., 2006). The precise analysis of the performance in the game, however, was very limited for a long tim due to the effort that was associated with game analysis normally carried out by observers sitting next to the field documenting the various actions of the players. The available software was dependent on the visual control of the observer. As actions and interactions in soccer, com-pared for example to tennis, are complex, it was difficult to bridge the gap between results of performance diagnostics in the lab and performance on the soccer field. Not long ago new technologies (AMSICO PRO) became available with which it is possible to identify, scan and track each player on the field for the whole game based on 2D-kinematics. Thus all actions and every movement can be assessed quantitatively. For an individual analysis of a player this is a remarkable advance as now relationships between physiological lab-and field parameters can be in-vestigated. Due to the above mentioned complexity of the game and the huge amount of digital data that is now avail-able (e.g. running distances, accelerations, running veloc-ity, shots, passes), traditional statistical procedures are no longer sufficient. In current research therefore the following strategy was mapped out. 1) Adjusting statistical tools for the analysis of complex data and systems (homogenisation, profiling, scale variation, local and global pattern detection); 2) Ap-plication to data of top European soccer teams (Englis