Can Relationship Analytics Predict the World Cup Winner?

At the time of writing, the Soccer World Cup 2018 Semi-finals have been completed and we are days away from a France-Croatia final and an England-Belgium play-off for 3rd place. Despite a plethora of statistical information available on team past performance, there was only one data set that could be considered representative of how players in a team related to each other; and that is the passing distribution information:

The passing distribution information records how players are connected and therefore related through the activity of passing the ball to each other.

Sounds obvious, but every other statistic focusses on individual performance, not how and whom players interact with. This is the classic “can a star team beat a team of stars?” question. And this is not confined to sporting events. In business it is common for us to focus on the individual performance, assuming that collaboration is simply the ‘sum of the parts’.Relationship analytics moves the focus from the individual to the relationship, in predicting performance. SWOOP Analytics  exposes relationship insights by looking at how people connect with each other as a social network at work. In this article we apply these same analytics to football teams, to try and predict who will emerge the winners in this weekend’s games.

Network Performance Framework

In a previous blog post we retrospectively analysed this same data for the 2014 World Cup finalists, identifying player roles with the SWOOP collaborative personas (Engagers, Catalysts, Responders, Broadcasters, Observers). This time we are using our collaboration performance framework, which SWOOP uses to benchmark collaborative performance through mining how staff interact using their enterprise social networking platforms.

Drawn from social network science, the framework identifies maximum performance when a network or team is able to maximise both cohesion and diversity amongst its members. For this exercise we measured these two dimensions using the passing distribution data drawn from the semi-final games, as follows:Cohesion = the number of reciprocal connections exhibitedUsing the passing matrix, we simply count the number of player to player connections that were reciprocated i.e. player A passed to Player B and vice versa.Diversity = the extent to which an over-reliance on a selected few players exists

We used our ‘key player risk’ measure, which calculates the number of players that are responsible for 50% of all passing moves. A lower number indicates a heavy reliance on a few key players, a clear risk. A higher number shows that the passing load has been more evenly spread across the team i.e. more diversity through more players being involved.In the example here, 3 players are responsible for nearly 50% of all passing movements.

Results

We analysed the 2 semi-final passing distributions with the following results:

We can see that in the France vs Belgium semi-final, France rated higher than Belgium on both Cohesion and Diversity and was also the winner. In the other semi-final between England and Croatia, the winner Croatia led England on Cohesion, but trailed England on Diversity. If we compare the relative performances, Croatia does come out a narrow winner. We therefore cautiously claim our performance model validated by these results.

Our Prediction

Well here it is, the ‘proof is in the pudding’.  If indeed our predictions prove to be true, we will humbly accept bragging rights. If we are wrong, we will hide behind the usual excuses of insufficient time to do a more comprehensive analysis i.e. the use of only a single prior game for history, and of course history is not always a good predictor of the future. But here goes!

The Final: France vs Croatia

From the results table we can see that France marginally trails Croatia on the Cohesion measure, but is comfortably ahead on Diversity. The network maps below identify the reciprocal connections as red lines and the larger circles identifying the key players. We can see that the reciprocal links are similar but it’s clear that Croatia has 3 passing hubs (Marcelo Brozovic, Ivan Rakitic and Luka Modric). France’s passing hubs are more distributed. This should help France, especially if they can shut down Croatia’s passing hubs.

Our Prediction: France by 1 goal

3rd Place Play-off: England vs Belgium

Referring again to the semi-final results table we can see that Belgium has a clear lead over England in Cohesion. On the other hand, England clearly leads Belgium on Diversity. It’s a toss of the coin. If we look at the proportionate differences, we get:

Our Prediction: England in a Penalty Shoot-out.

A Final Word

While this blog post is a bit of fun, there is a serious side. If indeed you believe that a star team can always out-perform a team of stars, in both a sporting and business context; we need to stop just measuring individual activity and performance and assuming team performance is simply the sum of the parts. We need to analyse what makes a team a team, and that is how they interact with each other. That’s what we need to measure and analyse when predicting future performance.