Soccer as a Network
Soccer analytics are not up to par with statistical analysis in other major sports such as baseball, basketball and American football. However, network theory provides a possible framework for modeling the most popular sport in the world.
The idea is to represent players as nodes and passes as edges in a network graph, with the thickness of edges indicating the number of passes between two nodes. The extent to which players are connected to their team can be measured by closeness centrality. This also allows us to measure how the path of the ball between players depends on another player based on betweenness centrality. This is a good tool for evaluating the balance of a team – if one player has extremely high betweenness then the team is highly dependent on them. For example, in the picture above, number 11 on Spain has the highest betweenness centrality. We can also measure which players are most important in an offense by using the PageRank algorithm. If someone receives a lot of passes from popular players, then they are likely to have the ball during a long possession.
I think there are certain ways this approach can be improved, for instance the location of each pass given to and by a player could be tracked somehow. Also, there should be some way of tracking the average and standard deviation of the length of chain of passes.