Using Network science to identify the best soccer players on a team
http://phys.org/news195911408.html#nRlv
In the above article, a team from Northwestern U. applied a statistical analysis algorithm to soccer players in order to quantify a player’s contribution to their team. Although this practice is common for most sports it is particularly difficult for soccer compared to basketball and american football because of the game’s continuous play and high amount of player-ball contact but low number of goals/assists. A key player for a team may be crucial for their team’s victory without actually recording a goal or assist. To account for this, the team modeled European national teams as network graphs where the strength of the edge connecting two nodes represented how often those two players passed to each other. In this way, key players who were often passed to or repeated combinations of passes between the same players over and over again could be quantified and evaluated visually.
In the example photos, three of Spain’s games in the 2008 Euro Cup are displayed. The Spanish team would go on to win the tournament that year and it is possible to see the strength of the Spanish side as a whole unit when represented by the graph. With almost twice as many total passes in their games as their opponents, the Spanish capitalized on smooth, accurate passing between all their players. In Spain’s 4-1 destruction of Russia, the graph clearly represents the inter-connectivity of Spain’s attackers, especially the Xavi and David Villa nodes which had strong edges to almost all their compatriots – the result of Spain’s efficiency in giving their best players the ball as often as possible. Compared to other teams, such as their Final match opponent, Germany, the Spanish network demonstrates a more distributed, secure structure. Germany’s plan of attack was readily visible from their network, where the defenders would constantly follow the same passing path to the front left nodes. Although this path was probably just as effective in moving the ball to the attack, the fewer edges among Germany’s forward nodes shows the greater distance separating their players.
Two of the network principles learned in class that are most readily visible here are the Strong Triadic Closure Property and the advantage of having more, strongly weighted edges between nodes when navigating along the network structure (as the soccer ball was passed between team mates). For most of Spain’s opponents, there were few strong links between nodes, most of the team was bound together with ‘weak’ links. Since these weak links are not likely to cause further links to form or strengthen, the opposing teams remained only loosely connected, and the distance separating some players was quite large. In comparison, Spain’s team, particularly when crossing across center field, is highly connected between all the neighboring nodes and these ‘strong’ connections likely caused greater cohesion among the team as a whole. The clique of fully connected players in the midfield gave the Spanish players more options and paths in which to move the ball from defense to attack. The second trait that is visible here as an advantage of a more stable, connected network is the fewer ‘jumps’ required to get from any one node to any other. Since the network represented here is the only set of paths that the ball can travel along to be in a potential scoring position, the ability to transition quickly among players is essential, both moving from the backfield to the opponents half, or within attackers once they have the ball in the opponents half. All these characteristics are represented in this network mapping of the soccer players, and helps to explain the ‘extra’ factor that comes from a team’s cohesion and unified structure – a network of connected nodes.
Hock. E Fan.
9/18/2012