ANALYSE THIS: Is there meaning behind few PL goals?

With 40 Premier League matches played, only 78 goals have been scored, an average of under two per match. Whilst it is reasonable to expect this average to increase, Infostrada’s head of analysis @simongleave has found potential reasons for the low scoring as this is now happening more frequently in general, and more often at the beginning of the season.

 

As someone with an extensive background in statistical analysis, my initial reaction to the narratives about this low goalscoring was one which I didn’t spend long enough thinking about. I simply assumed that the 40 matches we have seen so far are a statistical anomaly and regression to the mean will take us back to normal levels of goalscoring.

An article (link below) by blogger Mark Taylor in particular forced me to examine my position in more detail though. That analysis included a simulation which estimated the chance of so few goals being scored by this stage at 0.5%. I decided to replicate that but with real data rather than a simulation. Using all of the seasons from 1968/1969 onwards, and splitting the matches up into 40 match blocks, produces the distribution shown in figure 1 below. 

image

Click on the graph for full size version

Why 1968/1969 onwards? That was the season in which goalscoring dropped below three per match in the English top flight after every season throughout the whole of the 1950s and 1960s prior to that had had an average of above three. Goalscoring levels have never returned to three per match since.

The drop in 1968/1969 was also remarkable as 0.4 goals per match fewer were scored in 1968/1969 compared to a season earlier, a decline of 13% in goalscoring. Potential reasons for that are for another article.

The 45 seasons worth of top flight data from 1968/1969 onwards average 2.61 goals per match, which is comparable to the briefer Premier League era average of 2.63. Over those 45 and a bit seasons, there was only a 0.66% chance of 78 or fewer goals being scored in a 40 match span based on what actually happened. That is an extreme result and worthy of investigation to try and discover why.

In total, there are 126 spans of 40 successive matches in which 78 or fewer goals were scored and they occurred on 25 different occasions during the period under analysis. Can we make sense of the current low scoring by looking back at the other 24 occasions? They are listed in figure 2 below. 

image

So, the beginning of the season, represented in bold in figure 2, and the beginning of a calendar year, appear to be the two moments during the season in which low goalscoring is most likely to be observed.  My theory is that the bedding in of new players, particularly those from other countries and cultures, along with teams beginning seasons more conservatively are driving the frequent outliers that are now being observed in August, September and October. The fact that this phenomenon seems to begin in 1998 when foreign imports had really begun to increase and has only become more frequent since transfer windows have been introduced are good circumstantial evidence.

The occurrences at the beginning of the year could be due, in part, to the January transfer window, but also perhaps injuries and tiredness coming into play. Prior to the late-1990s, it was Febuary, March and April which had the dips in goals. Since then, this has moved to January, February and March.

So, finally, what should be expected this season? Clearly the scoring rate will not remain as low as 1.95 per match and there will be regression to the mean. It is quite reasonable to expect a decline in goalscoring after the last four years, three of which have been the highest scoring top flight seasons since 1967/1968. One reason for this could be the tactical changes influenced by Guardiola’s high pressing at FC Barcelona (thanks Sean Ingle), but teams eventually adjust to these tactical innovations and that, in turn leads to regression back to previously seen levels of, in this case, goalscoring.

However, a reduction in goalscoring this season will not necessarily be significant. In order to be reasonably sure that a change has taken place, as it certainly did in 1968, goalscoring needs to drop by about 7% in comparison to 2012/2013. If this season ends with 984 (2.59 per match) or fewer goals, significance will have been reached and we can say that something really has changed.

 

References:
Taylor, M. (2013), A Premier League goal drought. The Power of Goals, 16 September 2013. http://thepowerofgoals.blogspot.com/2013/09/a-premiership-goal-drought.html

Taylor, M. (2013), Where have all the strikers gone? The Power of Goals, 17 September 2013.

http://thepowerofgoals.blogspot.com/2013/09/where-have-all-strikers-gone.html

Trainor, C. (2013), Where have all the goals gone? Stats Bomb, 17 September 2013.

http://www.statsbomb.com/2013/09/where-have-all-the-goals-gone/

 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s