Every pass is grouped into 64 clusters using k-means clustering. The line plot of year-to-year difference measures the year-to-year difference of each cluster per team. The resulting distance score is normalized against all other year-to-year transition in the big 4 leagues in Europe. 50% means an average change of team’s style due to the average year-to-year turnover of players/coaches/tactics. 100% means the maximal change and 0% means the minimal change.
Most changes for these teams come when the team changes its coach.
The pass plot per team shows all of the pass clusters that are in the top 15th percentile over the years.
A detailed description can be found here.