Aplicando técnicas de aprendizado de máquina na previsão em tempo real de resultados de partidas de futebol: rocket e multirocket.

Abstract

Given the immense popularity of football and the associated betting market, an improvement in prediction accuracy has significant implications, both from a technical and economic perspective. With this in mind, the proposed project aims to explore, develop and evaluate several machine learning algorithms, including different neural network architectures, for the complex task of predicting football match outcomes in real time. By using a range of statistical variables (such as cards, shots on target, fouls, dangerous attacks, corners and goals) in a time series representing the state of the game, the project contributes to the advancement of the field of sports data analytics and has the potential to influence the sports betting market. This project is therefore not only academically relevant, but also has high commercial and social value, potentially influencing the way betting strategies are formulated and perhaps even how the game is played and analyzed.