In conventional game performance analyses, the investigators focus on data concerning the frequency with which techniques are used, the success and failure of performance and the behaviour patterns (e.g., Hughes, 1996; Hughes and Franks, 1997; Reilly et al., 1997; Spinks et al., 2002). However, in order to evaluate the overall team skill by considering game statements such as field positions of team mates and opponents, there is a need to develop a team skill scale which has causal structures according to the attacking or defending phase (Hughes, 2002). Additionally, the team skill scale should be able to evaluate multi-dimensional abilities of (a) the phases, and (b) objects of attacking or defending. Data based on such a scale will provide coaches with useful information for designing training programmes. Moreover, conventional game performance analyses take too long to return the information to players. However, the automatic measurement of the location of players in games owing to advances in motion analysis techniques (Taki et al., 1996) will be able to decrease the time required to return the information to players and to reduce the hard work of analysts in collecting data.