Hi /sci/, midwit needing some help here!

If you're familiar with team based games, it would be helpful.

A game consists of 10 players, 5 on each team. Each player picks a champion.

Champions have unique characteristics, some are aggressive, some are supportive etc.

How do I go about modelling how much 1 point of X characteristic has against another characteristic?

Let's say that composition A has 5 point attack & engage, and composition B has 4 points in counter-engage and 3 in protect.

Which method should I use to determine the effect (I have around 200 matches), and how do I even set up the data for such?

I was thinking of logistic regression and trying to predict win (1/0), but how can it distinguish which teams win against another?

If I put the team that won and lost on different rows, the individual characteristics will wrongly weighed as the game works more like a rock paper scissors. So maybe protect loses against almost everything except attack, which it has 90% winrate against. If I split the rows, protect will just always be negatively correlated with winning.

If you need visualization / more clarification please do ask, I really want to do this model.

Thanks!

Pic not related, but I am in a thinker pose too.