Network Science reveals the secrets of FC Barcelona's total football team
According to a recent study, the 2009/10 FC Barcelona squad's performance is beyond any other football club in the world. Under the eloquent leadership of Pep Guardiola, the team re-wrote the football history by winning six main competitions, including LaLiga and the most prominent contest – the UEFA Champions League.
There is no other team that managed to collect so many achievements in such a short time. Guardiola taught his team a unique style of football, which remained iconic to this day for Barcelona. Their goal is to keep possession of the ball by making short and quick passes to the closest player. If possession is lost, the players should immediately aim to recoup the ball.
This efficient gameplay strategy is commonly known as total football, or as Barcelona’s fans call it, the tiki-taka. While this strategy can be described in broad terms, football analysts are more concerned about capturing it through the data collected by research companies.
By comparing conventional metrics (goals, passes, points, and shots), Barcelona outshines its competitors. However, these metrics are not enough to determine the team’s playstyle. So, sports researchers would love to find a way to explain the considerable performance difference between Barca and the other football teams.
Javier Buldu at the Universidad Rey Juan Carlos from Spain successfully captured the gameplay style by combining different metrics to extract the signature style of Guardiola’s Barcelona.
Network science in sports is relatively new. Their idea is to represent each football player as a node and create a link between them every time they make a pass.
The connection becomes stronger as the number of passes increases. The data also includes the position on the field of each athlete when a pass is made. By the end of the game, the network becomes an authoritative record of statistical relationships between players and how the game evolved.
But the possibilities with network science are endless. Analysts used it to study the spread of disease, the Internet, forest fires, or even the probability of war. Marketers use network science to research the consumer’s behavior towards promotions for new gamblers or discount coupons.
Various MIT researchers have already used this approach to determine football team performance. After running countless tests, they discovered that the networks would form small worlds (in essence, networks can be crossed in a much lower number of links than there are players in the team). Another exciting discovery is that individual players are more central, meaning that the ball is more likely to be passed to and from them. Scientists have also identified that specific playstyle patterns or motifs are relatively common, like passing the ball between three players and forming a triangle.
Buldu and his team took this approach to the next level. Instead of analyzing the network from the entire game, they studied the way it changes throughout each match individually, by generating a network created from the first 50 passes. Then, they used a sliding window to inspect how the system changes as the game progresses. Essentially, their approach was adding the 51st pass to the network while removing the first one, and so on.
First of all, they generate the passing network for both teams in every La Liga game from the 2009-10 season(380 total games played between 20 top-tier football teams). Then, they calculate several network features that include the clustering coefficient, which describes the performance of player triplets passing between each other. Barcelona’s clustering coefficient was much higher than any other competitor.
Scientists also discovered that the average shortest path through the network (which describes how efficient the ball is passed across team members) is quicker for Barca than any other team. The largest eigenvalue of the connectivity matrix, which measures the network’s strength is much higher for Barcelona than any of the other football groups.
After analyzing the network’s evolution over time using the 50-pass technique, Buldu and his team identified the metrics that enhance the probability of scoring or receiving a goal, showing that Guardiola’s organization of Barcelona is different from the rest.
For instance, this study reveals how Barcelona plays further up the pitch than most other teams, by measuring its average position on the field (a.k.a. team’s centroid)
Barcelona also has the highest centrality of a single player in Xavi, who is well-known to be one of the best midfielders in football history.
Of course, the analysis also revealed some weaknesses. The expansion of the dispersion of players around the centroid increases the likelihood that Barcelona will concede a goal. In simpler terms, when the players are spread out, the enemy team has a higher chance of scoring a goal.
While these insights uncovered many patterns in Barcelona’s gameplay style, it’s nearly impossible for the competition to copy it for success. Identifying trends is one thing, but putting them in practice is quite another. Without any doubt, network science will become indisposable for football analysis in the future.
This post first appeared on www.soccerex.com with a different title
