A Hybrid Approach To Build Automatic Team Composition In League of Legends

League of Legends is one of the most played games available in the market. The game had over 1.2 billion play hours from June 2011 thru July 2012, almost the double of the second place. We can associate the success of this game with the growing eSports Market. In 2017, during the Mid-Season Invitational event, a Riot Games event held in Brazil, over 364 million viewers watched the tournament. To reach success in this sport category, strategy and a good knowledge of the game is very important. Also, it is expected that the team mates and the coach define a good strategy while picking the characters, also called champions, which the team is going to play with. Our goal in this paper is to model a expert system to help the players decide which champion should be picked assuming that the champion is updated in the metagame and has a good synergy with the other champions. For that, the MinMax algorithm was used to model and create all possibilities of choosing champions, and also a classification algorithm to evaluate each option and give them a grade to optimize the MinMax algorithm through alphabeta pruning. The proposed modelling showed itself promising. Its application on eSports may result in the development of team compositions never thought before in which may be very beneficial to the players.