Who should play in which position in soccer? Empirical evidence and unconventional modelling

Because of the many variables, team games are one of the most complex systems one can find in sport. For this reason, deterministic modelling is not appropriate in order to account for the complexity and uncertainty of sport games. Computer scientists have developed several alternative models in order to deal with these complex, uncertain and non-deterministic problems. Fuzzy logic seems to be the appropriate modelling paradigm because this approach allows to deal with diffuse concepts. When coaches are asked which properties qualify players for specific tactical positions they usually operate with such diffuse concepts, e.g., good risk taking or heading capabilities for strikers. Furthermore, a coach may apply some rules for assigning tactical position, e.g. ‘if a player is an excellent goal-getter or he is at the same time a moderate goal-getter, has a good heading capability and is a good risk taker, then he or she should play as a striker’. This kind of reasoning reflects exactly the logic structure of the decision rules proposed by fuzzy logic.For this reason, we interviewed soccer coaches and then applied the concepts of fuzzy logic in order to develop a simulation program that gives recommendations for tactical positions of players, depending on their particular profiles of features. We interviewed 14 coaches and asked them questions concerning the properties and observable criteria of the following tactical positions: - Goalkeeper- Sweeper- Central defender- Midfielders (central vs. wings; offensive vs. defensive)- StrikerA questionnaire based on the results of previous studies was developed. This questionnaire consists of 48 items addressing the following aspects: - 9 items concerning the education and basic tactical concepts of the coach- 7 items concerning the properties required for a specific tactical position (pre-arranged items with a 1-to7 agree-disagree scale and the possibility to add items)- 28 items concerning the observable criteria for the properties (open questions)- 3 items concerning the subjective difficulties when completing the questionnaire (1- to-5 very difficult-very easy scale)- 1 open item for comments and recommendations.Based on the concepts of fuzzy logic a computer program was developed in order to simulate the determination of appropriate tactical positions depending on the players’ profile. In the following text this program and its components are introduced.As a first step we differentiated three types of positions (i.e., striker, ‘build-up’ player organising or conducting the offensive play, defensive player) and established the following rules for determining specific positions: - Typical features for a striker are goal-getter qualities, header capabilities, dribbling capabilities and 1 to 1 play.- Features that are relevant to ‘build-up’ players are view, creativity and availability for team mates.- The respective features for the defensive player are covering behaviour, defensive power and hardness.This model is very simple and is just a first approach to the problem; it is possible to change the relevant features and to add new features and tactical positions or both. So the model can be freely adapted to the needs of the users. The proposed interactive program can be used both for simulation and assessment; this may help to improve our knowledge concerning this exciting, and ever popular sport.