The Social Network Analysis of Football Matches

  • "Football is a simple game based on the giving and taking of passes, of controlling the ball and of making yourself available to receive a pass. It is terribly simple." -- Bill Shankly
 
  • "Football is simple. But nothing is more difficult than playing simple football." -- Johan Cruyff

Did you know that several academics have used network analysis in an attempt to derive insights into the beautiful game?

During the 2010 World Cup, FIFA analysts compiled data listing every pass exchanged between players, and put it online (unfortunately it seems to have disappeared since then). This data was used by researchers at Queen Mary University of London to create a network theory explanation of how each team played. There's also a media-friendly summary of their work, and a more detailed write-up in the following paper:  

And the QMUL group aren’t the only researchers seeking to apply network analysis to football, there’s also the long-running ARSFutbol research group at the university of Buenos Aires. Some of their work looks very interesting, it's in Spanish, but Google Translate will help you read it in your own language.

I also found another paper analysing the 2010 World Cup from some Spanish researchers:

As this BBC article explains, most big clubs now have a team of data analysts, and companies like Prozone produce detailed data on every aspect of what happens on the pitch. Unfortunately clubs pay a lot of money for Prozone data, so it's unlikely to be publicly available; so if you're hoping to do some bedroom-based analysis, you might be out of luck.

Not all are convinced of the value of this type of analysis, mind you. Tottenham manager Andre Villas-Boas said: “I have never used Prozone. I don't use it because I don't believe in it". Whereas other managers, like Rafa Benitez, are more keen on the data-driven approach.

That network theory might be applied to football is not particularly surprising. After all, football is a passing game, and interactions between players are fundamental to a successful team. Hence when we fans watch players playing well, we frequently call them “influential”. And network theory has proven to be especially good in determining who or what is influential amongst the millions of documents now online.

By comparison with the Byzantine complexity of online content, analysing what happens on a football pitch seems child's play. The graph of a football match will only have 22 players (nodes), a handful more if you include substitutes.

However, to paraphrase Cruyff: calculating the numbers is simple, but deriving useful insights from them - that's difficult. For instance, say you notice a lot of passes are going through your centre-forward (he has high betweenness in the network jargon), is that a good thing? Was that part of your team's plan? Or should you be yelling at your centre-forward to get into the bloody box and to start causing some fecking mayhem?
 
Does network analysis have any value to our understanding of football? Will it one day be able to quantify what football professionals have come to understand intuitively - demystifying it for the rest of us? Or perhaps the spirit of football will always defy analysis, generating as many opinions as there are fans watching it. After all, what model could ever hope to explain the Miracle of Istanbul...

0 comments:

Post a Comment

Twitter FeedRecent Twitterings

    Do we share an interest?

    c o n t a c t m e @ j a r o n c o l l i s . c o m