Predicting Football Using R

I recently gave a presentation to the Manchester R Users’ Group discussing how to predict football results using R. My presentation gave a brief overview of how to create a naive Poisson model in R and apply the Dixon and Coles adjustment to it to account for dependance in the scores. The slides are below [...] Read more ›


Is there a relationship between “mean-time-between” metrics and effective time?

Since I started to apply mean time between fouls and mean time between stoppages to World Cup data, I was curious to find out if there was any correlation between these “mean-time-between” metrics and the effective match time.  I’ll examine the correlation issues in this post. For reference, here is my post on MTBF for […] Read more ›

Looking at MTBS for the 2014 World Cup

Continuing our application of “mean time between” metrics to the World Cup, I consider mean time between stoppages. These metrics, and all those related to the 2014 FIFA World Cup, are derived from data from Press Association‘s Match Story feed, which was distributed through the Soccermetrics API. Below is the chart of average of mean […] Read more ›

A first look at MTBF for the 2014 World Cup

In my previous post I introduced the concept of mean time between fouls and mean time between stoppages.  In this post I calculate mean time between fouls over all matches of the 2014 FIFA World Cup. The data are taken from Press Association’s Match Story feed of the World Cup, which is ingested into the Soccermetrics […] Read more ›

Introducing mean time between fouls and stoppages

I was looking through this blog at the number of posts that I’ve written about referee performance, and it turns out that there are, as of this date, five posts under that category.  Three of the articles are reviews of research publications, so I’ve actually written very little about the subject.  That’s about to change […] Read more ›

Subbing Your Striker.

Substitutes provide a fascinating look into the changing dynamics of a football game.Scoring accelerates as the game progresses and in this post from 2012 I looked at the tag of “super sub” that had become attached to Edin Dzeko and how it owed much to… Read more ›

Your Biggest Opponent Might Be Your Teammate.

In attempting to answer the question “how good is a striker?”, the focus unsurprisingly falls on their goal scoring abilities. Raw goal scoring totals have been replaced by such statistics as “goals per 90 minutes” to reflect playing opportunity, often with corrections to remove penalty kicks and to allow for the different scoring environments that are present later in a game.

Latterly, actual goals have been replaced by expected goals, based on the likelihood that an average player would score with a shot or header from a particular location on the pitch. 
The ability to get into good scoring positions may be as important as the successful execution of a chance.

Random variation may in the short term produce actual goal tallies that flatter or under value a player and so his “virtual” goal per minute numbers may be a better indicator of his likely long term scoring potential.

The difference in talent levels at the very top level of finishing is likely to be very small and the ability to outwit the defence and fashion a chance may be more important than short term goal scoring achievements.

The opposing defence is quite naturally considered the natural opponent of a striker trying to amass expected goal value, but often there is also another factor that might cause his numbers to rise or fall.

Liverpool was fortunate to be able to call on two exceptional strikers, last season. Both Suarez and Sturridge reached impressive, respective goal totals, both real and expected. In addition, both missed playing time for a variety of reasons. But Liverpool was able to call on at least one of the pair for every Premiership game.

Suarez’s suspension left the striking stage clear for Sturridge at the start of the season and then the roles were then reversed when the latter missed a run of matches from early November to mid-January.

While in tandem, the two strikers could possibly be competing for the best chances created by their teammates, while in the other’s absence, each may have laid claim to the majority of the prime opportunities.

Therefore, I looked at each goal attempt made by Suarez and Sturridge over the 2013/14 Premiership season as defined by shot location and type. And, reflecting the wider picture, these primary parameters were significant indicators of whether a goal was likely or not to be scored.

I then added a variable to differentiate between when both Suarez and Sturridge were on the pitch and when just one of the strikers was present. This new variable was also a significant indicator of likely success, reducing the likelihood of a goal being scored when the duo were playing together.

In the case of Sturridge, his expected goal per game number falls from 0.9 when he flies solo to 0.67 when he partnered Suarez. A fall of nearly 30%.

The same is seen for Suarez, 1.1 expected goals per game alone and 0.75 when paired with Sturridge. Once again, this is a fall of around 30%.

We are looking at nearly 300 total attempts, but the results may just be a quirk of this dataset. However, on a shot by shot basis, it does appear that the players are taking a bigger proportion of low expectation attempts and fighting over the prime cuts when both are playing.

Suarez’s average goal expectation per attempt fell from 0.23 to 0.14 when Sturridge joined him on the field. Sturridge’s comparative figures went from 0.27 to 0.18 per attempt.

This is a single case involving two extremely high class finishers and there are at least eight other outfielders to consider, but it perhaps seems that the pair as a duo may have been shooting sometimes, when the position was too heavily defended.

This may have had implications for Liverpool’s strategy when both played, although it is now a moot point.

Perhaps more pertinently, each player appears to depress the expected goal record of the other by their considerable presence on the field. And this may have implications for player assessment as well as projection at future clubs, where the most difficult opponent to overcome in posting impressive scoring statistics may well be your teammate. 
Luis Suarez debuts for Barcelona in the near future. 

Read more ›