The structural features of a soccer team's passing network are indicative of overall play style but may also indicate a higher probability of winning a match, according to a new study that is the first to use passing network indicators to model the probability of winning a match.
Given that passing represents more than 80% of the events in soccer, Italian researchers were interested in seeing whether a team's passing patterns impact the probability of winning a match — and it turns out they do, with certain features of a soccer team's passing network increasing its probability of winning by one-third.
Additionally, looking at traditional measures for calculating odds of a win, researchers found that for each shot attempt at the goal, the probability of winning the match rises by 9.6%, which indicates that "shots, aggressive attitudes and the amount of running can be crucial in a high-quality team's tactics and generally, increasing the probability of winning the game," according to the study, published March 29 in Knowledge-Based Systems.
Often when studying the probability of winning a soccer match, statisticians examine a team's shots, shots on target and ball possession statistics. Researchers have studied these facets in novel contexts, such as a study published in March that found that the absence of spectators was associated with a significant decrease in home teams' shots and shots on target.
This study, however, is among the first to employ a passing network analysis that takes into account centralization, clustering and cliques from a soccer perspective, in addition to the relationship between players in passing and the type of interactions among them at both a micro-level, studying the individual players, and a macro-level, studying the team as a whole.
By doing this, the researchers were able to analyze the structural features of passing networks and evaluated teams' style of play, presenting their findings in classic network analysis graphic visualizations to compare teams and their individual level of connection.
Lucio Palazzo, a postdoctoral researcher at the University of Naples Federico II and co-author of the study, said the idea for the paper came from a discussion between Palazzo and co-author Riccardo Ievoli, a postdoctoral researcher at the University of Ferrara.
"Me and Riccardo [were] arguing about the idea that cohesion between players in terms of passing could be in a certain way useful to determine the results of a match," Palazzo told The Academic Times.
Using data from 96 matches in the Group Stage of the 2016-2017 Union of European Football Associations Champions League, which involves the 32 best European teams, the researchers modeled the probability of winning the game based on network passing indicators.
The passing network indicator with the strongest effect on the probability of winning a match was "betweenness centralization," according to the researchers, which is considered to be one of the most reliable measures in network analysis.
At its simplest, betweenness centrality is based on the number of paths through a given node in a network. In terms of soccer, this means that individual players with higher betweenness centrality scores are those who are more influential in terms of passes among other players — the ball comes to them more often than other players and they pass the ball more often than other players, according to the researchers.
These players with highest betweenness centrality act as mediators or bridges, playing an important role in passing the ball to other players. Palazzo and his colleagues noted that central midfielders and defenders are expected to exhibit these qualities during the match and are therefore expected to have the highest betweenness scores on the team
For each additional point of betweenness centralization on the team, the probability of winning the match increases by 35.8%, the researchers wrote.
The overall passing properties of a team are also indicative of play style and can help predict a team's likelihood of winning. For example, in their network analysis of a game between Barcelona and Borussia Mönchengladbach, the researchers saw Barcelona's well-known "tiki-taka" style displayed in the graphical depiction of the passing network. "Tiki-taka," developed by Barcelona and eventually adopted by Spain's national soccer team, is a ball possession strategy using a series of short, rapid passes.
In this particular match, Barcelona completed 939 passes with approximately 92% accuracy, and "tiki-taka" is associated with a higher probability of winning a match, according to Ievoli, Palazzo, and Ragozini.
The model employed by the researchers in the study has a high degree of value and provides useful tools in terms of the practical side of soccer, the researchers wrote, as the information on passing networks can provide real-world, useful information for practitioners, trainers and experts as an easy way to detect patterns, evaluate ties among players and their positions in the team's lineup, and reveal team tactics and connections between positions.
"The measure can be difficult, but the concept is simple," Ragozini said.
The study "On the use of passing network indicators to predict football outcomes," published March 29 in Knowledge-Based Systems, was authored by Riccardo Ievoli, University of Ferrara; and Lucio Palazzo and Giancarlo Ragozini, University of Naples Federico II.