hello Benoit, I am a passionate like you on data (fifa) and football and I am trying to define a way to measure distance between teams. it seems easy but it’s not in my opinion. I wonder if you have tried something like that and if you have any pointer. On my side I tried the easy way by averaging the players in every team and the using PCA which gives good results but I think this is suboptimal and there could be better ways by comparing player by player for every role in a team. this should be more precise than flattening by using the mean. but in this case there is the problem that every team has different number of players, so I should first uniform the teams by choosing the top for example 14 according some rule or something like that. Then I should compare role by role every player but even in this case not all the teams have same number of players in a role and also it depends on modules any team plays which is not easy to define as there could be many modules… Anyway have you any thought about a situation of this kind? (I’m on Medium too) regards Andrea