Modelagem estatística para a determinação de resultados de dados esportivos

Abstract

The basic result of a soccer game is the final scoreboard, which can be seen as a bivariate random vector. Theoretically and based on existent literature we can argue that the number of marked gols by a team in a game obeys a (univariate) Poisson distribution. Thus, the Bivariate Poisson distributions are studied, in special for the class "of Holgate" (1964). Using as information the recent results of the teams, whose confrontation we want to model, several methods were used for parameters estimation of the Bivariate Poisson class "of Holgate". The idea is to use procedures that supply the probabilities of occurrence of placares, so that thus the probability of the occurrence of a certain result (team home´s victory, draw or defeat) can be calculated properly. The parameters of Bivariate Poisson distribution "of Holgate" are assumed to have a dependence factors, such as attack, defense and field, that possibly explain the numbers of goals.