Nowadays football betting is becoming more and more popular around the world. In the last few years several works have been done and improved in order to develop models able to predict the outcome of football matches. In our analysis we study the Dixon-Coles model for the full-time scores and then we focus our attention on the difference of goals, since it seems to be more advantageous than modelling the scores themselves. We develop two basic models for the difference of goals based on the discrete Normal distribution that gives us interesting results as compared to the Skellam distribution. Furthermore we study the Dixon-Robinson model for the goal times and we investigate the possible clustering of goal times data. Using self-exciting point processes, we found that the scoring rate in a football match tends to be higher during the minutes straight after a goal has been scored. Our general results want to be a solid starting point for more sophisticated analyses.