Location systems in game sports provide a wealth of data, which is capable of reconstructing game play from a tactical point of view. Precise position measurements up to 50 times per second/player during an entire soccer game enable in depth analyses of individual and collective game behavior. Small sided games (SSGs) are used in soccer training to simulate specific game situations. By the given constraints (smaller pitch size, lower number of players) SSGs are used to provoke intended player behavior with a higher rate than it occurs in the full size game. Due to these properties SSGs are also suited to analyse successful/unsuccessful tactical patterns in a reasonable framework. The research described in this paper aims to identify behavioral patterns in soccer SSGs based on the players` positional data. The approach is demonstrated by a 3vs2-SSG (3 attackers playing versus 2 defenders), which is often used in soccer training to acquire collective and individual goal scoring skills against a shorthanded opponent. Based on expert information the SSG is temporally segmented and for each defined key moment and temporal phase a couple of parameters are derived from the position data. Statistical methods (e.g., the laplacian score for feature selection) reduce the high dimensional parameter space to a manageable number of variables. This resulted in the 3vs2 analyses in 9 parameters describing the shot and the assist (last past) situations. For each of the two situations unsupervised pattern recognition was performed by spectral clustering. For the shot as well as for the assist situation three different patterns were identified. The goal ratios for the assist patterns range from 0 to 29%, the goal ratios for the shot patterns from 8 to 24%. By a fuzzification of the quantitative parameters into linguistic variables the outcome of the pattern recognition process becomes understandable and interpretable for experts like coaches. The results give evidence that the most efficient assist is a middle distance to goal pass from the side to the centre, whereas rear-faced passes are far more successful than forward-faced passes (47 to 8% goal rate). Shots with the highest success rate are performed from a middle distance to the goal in the centre of the field. The findings of our study give data based hints for successful tactical principles in soccer, which are likely valid also for the 11-a-side game. For the 3vs2 test case we can support the common coaches` assumption that ball possession at the wing moves the defenders to the side, which creates space in the centre field. This enables opportunities with the highest goal scoring probability.
© Copyright 2016 21st Annual Congress of the European College of Sport Science (ECSS), Vienna, 6. -9. July 2016. Veröffentlicht von University of Vienna. Alle Rechte vorbehalten. / Übersetzt mit https://www.DeepL.com/Translator