malricI understood how the player rating takes place.
Instead, what is not clear to me is whether the algorithm has been given the opportunity to give a historical evaluation of the player.
For example, a good casual player, not 1v1, plays in a wellcome team, and as often happens, through no fault of his own, (newbies who build detriments after 1 min, etc.) lose more than they win, even losing 10 games in a row . The elo which initially varies slightly, as I have seen from personal experience suddenly collapses with the succession of defeats.
Drops in rank. He is no longer purple. The rating has dropped.
But does it really prove its strength?
In my opinion not. And this affects the balance of the teams. So if his Elo dropped from 2700 to 2000, in a team he is considered as a 2000 player, but his effective strength is definitely higher.
And I think it affects the balance.
That's why I say that rank should be considered as an additional variable.
If you're a chess GM, they don't take your title away if you lose 20 games. You are always a strong player. This can't be measured with a number, but with rank it is. If you have reached that rank, it means that you are skilled enough.
So in summary, in the division into teams, the balance should take into account when dividing the two teams:
1. first taking into account the ranks of the players,
2. then calculating the rating in the next phase for the remaining players where the ranks are not evenly divisible between the two teams.