October 12, 2010

Measuring Individual Performance in a Team Context

Let’s do a quick test:

Name one player from Manchester United.
Was it Wayne Rooney?

And one for Barcelona.
Was it Lionel Messi?

And, finally, how about a player from the Socceroos?
Was it Harry Kewell?

You get the point. Teams almost always have a ‘face,’
typically belonging to a gifted player that’s able to impress the crowds with skill and attacking flair. These are the players that win all the accolades, sell more merchandise to the spectators and end up with the multi-million dollar contracts.
But one player does not a team make. At the end of the day, soccer is a team sport – and success on the pitch depends on harmonious, efficient, and effective interaction among the players on the team. Each player contributes.

Yes, you say, but isn’t Wayne Rooney undeniably the most important player in his team? I mean, surely Manchester United wouldn’t have paid 25 million pounds for less than the best … right?

The reality is that it’s not easy to tell. Owners, coaches and trainers would all love to know how much each player in the team is contributing to the team’s overall success. And in fact, so would most business-owners. The question of how the individual contributes to the whole is an important one, for a lot of people.

So it’s actually quite surprising that scientists haven’t come up with a reliable way of identifying relative contributions in a team environment.

Why is it so difficult? Soccer, specifically, moves quickly and involves near-continuous flow of the ball between players. It’s a complex game, to say the least. Measuring individual contributions in soccer would be like … measuring how influential a particular person was in the spreading of a rumour at work. Or figuring out what other organisms might suffer if sharks became extinct.

But … wait! There is actually a branch of science that deals with this – it’s called Network Theory. Network theory looks at the associations between individuals and is used most often in social and ecological contexts. But some scientists are beginning to adapt this theory to soccer performance – and with very interesting results.

Network theory in soccer
Recently, a group of researchers tried to assess the performance of individual players in a soccer team by developing a new methodology based on Network Theory (Duche and colleagues, 2010). The guiding principle for the work was that a soccer team must move the ball towards the opponent’s goal by keeping possession long enough to create shooting opportunities.

Four measures of individual performance were based on this logical requirement for football success:
(i)    individual passing accuracy (% of passes that were on target),
(ii)  individual shooting accuracy (% of shots that were on target),
(iii) player flow centrality, and
(iv) pass flow centrality.

Player flow centrality refers to the percentage of times that a player is involved in the paths of passes that result in a shot - clearly, the better-performing players are more likely to be involved in paths of possession that result in shots. Finally, pass flow centrality measures the involvement of a player in any paths of possession.

(If you’re finding this a little hard to picture, I recommend you check out this article on Network Theory in ecological systems, written by New York Times writer Carl Zimmer.)

The Results
Based on the above methodology, Duche and colleagues (2010) assessed the performance of all the individuals that played in the 2008 European Championships competition. Overall, their highest-rating players were also those that were rated highly by the general sporting community. In fact, 8 of their top 20 performers were selected in the 20-player UEFA Team of the Tournament. Duche and colleagues’ top 2 rated players in the tournament were Sergio Ramos and Xavi Hernandez, both on the Spanish team that won the tournament.

Where do I think we go from here?
Well, I’ll talk more about this another time - our soccerscience.net group has done our own analysis of individuals in matches - but I think the method that Duche and colleagues have developed is a little too-heavily weighted towards attacking, rather than defensive, performance. We can’t forget about the importance of an individual’s ability to restrict (and constrict) an opponent team’s possession and opportunities.

However, this really is only a minor critique given the advance that this study has provided to the scientific study of this sport. I think this work is very exciting – and has yielded promising results. So it’s even more exciting to see what will come next!

Written by Dr Robbie Wilson with Dr Amanda Niehaus,
www.soccerscience.net

No comments:

Post a Comment