January 11, 2013
Division wide goal fest or just a few freak results?
It seems as if goals have been flying in at all rates throughout the current Premier League season, particularly over the festive period. With the hiatus provided by the F.A. Cup 3rd round ties, now appears to be a good opportunity to review the number of goals scored this season in a statistical light. Firstly, goals have been scored at a greater rate per game, on average, since the start of the campaign. The last two completed seasons have each produced an average of 2.81 goals per game. The current rate is running at 2.87, so clearly the end of season total is on target to post a three year high.

Some results stand out in the memory, particularly Chelsea’s 8-0 demolition of Aston Villa and Arsenal’s 7-3 success over Newcastle. The question then becomes whether these results are in line with the goings on from the rest of the season’s fixtures. Singularly, these two seemingly freak results are not out of place when compared to the rest of the data set. Using the Poisson distribution as a guide, it was well within expectation that we would see 2 games producing 8 or more goals by this point in the season. The fact that there have only been two such results shows that the data has not been skewed by just a handful of high scoring games. Indeed, it can be expected that there will be 2 more games played that yield 8 or more games before the conclusion of the fixture list.
More surprising is the fact that there have been more goalless games than may have been statistically expected. Indeed, there have been nearly 50 % more 0-0 games than the model predicted. 3 goal games have been the most commonplace outcome and there is little reason why this should not continue. The most exciting revelation from this observation is that teams seem to be playing for a result once the scoring has been opened and as the season draws on this more likely to be the case. Overall, the data set is overly represented at both extremes and at the integer value nearest the mean. Intuitively, it appears that any biases have cancelled each other out and there is probably value in using this data to help make predictions for the remainder of the season.
The contribution of the teams placed first and second in the table warrant comment. For all of Manchester United’s free scoring it has been highly surprising that they are yet to score more than 4 goals in a single league game. Given their goals per game average and the quality of their forwards it would have been statistically more probable than not that they would have hammered an opponent by now. It is probably only a matter of time and is one reason as to why the goals per game rate is likely to be maintained at nearly 3. Manchester City are yet to really hit full stride and as a result have probably underperformed in front of goal. Aguero has had a truncated campaign while Balotelli hasn’t fired at all. Again, any improvement from City will also help to maintain the league’s scoring rate. As things stand they have not produced a skewing effect on the data and further reiterates that it is likely that the high scoring rate is a division wide phenomenon.