Alternative count regression models for modeling football outcomes

In this work, we propose the use of discrete counterparts of the Weibull distribution along with a copula function for modeling football results, as an alternative to existing bivariate Poisson regression models and extensions thereof. We expect that the choice of the marginal distribution and dependence structure, which try to capture known features of the data, can be beneficial in terms of fitting of the developed models; to check this conjecture, an application to the Italian Serie A championship is provided.