Previsão da eficácia ofensiva do futebol profissional: Um caso Português

The forecast plays an important role in the planning, the decision-making and control in any domain of activity, including the sportive phenomenon of the soccer. The experience has shown that the extrapolative or not casual models (univariate models), that use only the information of its past values to forecast the future, can often predict future with more accuracy than causal or multivariate models. In this paper, we model and forecast the offensive effectiveness of the soccer team Sport Lisbon and Benfica, in Portuguese soccer league, by using deterministic methods (linear trend, moving average, exponential smoothing, holt, naïve) and stochastic models (ARMA models, random walk). The model selection criteria used in our study were the mean squared error, the mean absolute error and the mean absolute percentage error based in a one-step forecast of the last three observations.