A Framework Based on Optimization and Time Series for Team Composition in Cartola-FC

With the growing popularity of \emph{Fantasy Games} such as Cartola FC, there is a need for efficient strategies for composing virtual teams. In this work, we adopt methods of prediction and analysis of historical data from the Brazilian Championship with the aim of maximizing team scores, reducing virtual costs and avoiding choices that compromise performance. We propose the use of different predictive methods to anticipate the future performance of players, applying optimization techniques to select the best possible formation, considering cost and position constraints. The results showed that the ARMA model was more accurate in predicting player performance than other prediction methods. This systematic approach allows users to make more informed strategic decisions, maximizing the performance of their teams on Cartola FC.