Invasion team sports constitute a class of sports in which competition is based on the opposition between two teams, in a common space, with the aim of scoring through individual, group and collective actions. Recent methodological and technological contributions over the study of the match dynamics in these sports are limited by the lack of formalization of the theoretical assumptions applied. The absence of a formal language to support these studies induce the imprecision in the description and, consequently, in the analysis of players' behaviors in the confront, leading to the loss of relevant information. A main element of a formal model of match dynamics is the strategy since the optimization of performance in these sports depends on the collective action and in order to it occurs it is necessary that the team has a strategic plan. This dependence defines a fundamental hypothesis, according to which the team that prevails for a longer period playing according to its own strategy has advantage. Through an inter-disciplinary approach between Physical Education and Computer Science, this research investigates the complex problem of the planning and the analysis of players' behaviors, based on a formal model and innovative computational solutions for data acquisition and evaluation. The study of the hypothesis over the reasons for the success or failure in the match may result in contributions to the understanding of the structure and use of the team strategies, besides the development of a methodology for data processing and analysis related to the players' behaviors in the confront. These findings may also contribute to overcome the scientific challenge of simulating the match. The expected results from each of the four phases of the research and the type of approach applied to each of them are described in the sequence. First, the match phases (e.g., offense, defense, etc.) will be modeled and their main properties will be defined. The match phases formalization may contribute to the improvement of methodologies for the design and analysis of strategies. Second, based on the defined structure of match phases, it will be developed a methodology for inference of the players' behaviors in the confront, supporting a representation of the analyzed strategies. The methodological problem to be overcome in this phase is the differentiation between executed strategy and random behaviors through the identification of stable patterns of behavior. Third, based on the patterns of behaviors inference there will be defined a set of criteria to the complexity evaluation of reconstructed strategies. The representation through graphs allows performing topological analysis for studying the strategy complexity and strategies comparisons, respectively, of the same sport and between different sports. Fourth, the creation of a software for designing and analysis of strategies. The software will be supported by the defined structure in the previous phases of the research and should be useful for supporting coaches to design their strategies interactively, limiting frequent errors. Besides the support to the investigation of the central hypothesis of the study, the expected results should also have some spin-offs: i) design of video-games with levels of strategic realism not observed in the actual games; ii) use of the strategy representation defined as prior information for automatic recognition of collective movements of a team based on digital images of video recorded matches; iii) applications to artificial intelligence studies on the cooperative relations in competitive environments (e.g., robots soccer World Cup).
News published in Agência FAPESP Newsletter about the scholarship: