Talk:Matchmaking/@comment-78.14.125.209-20100731043126/@comment-3014924-20100922090537

That's not totally true:

"1 has 100% to win the game vs any of team B 0 has 0% to win vs any of team A " => It's not calculated that way but anyway...

"Result that team A will have 1 good player 1 decent and 3 noob Team B will have 1 noob and 4 good player."

=>That's theorically possible, however, with more than 10 000 users, you would have to be in either extremely good or bad to play with a lot of ELO difference. Usually, the full team will be same ELO class.

The only real cases this could happen is premade. Which they worked with their modifiers. It's statistically easy to verify whether the system is correct or not with tests on the Mean Value. You could even look at stats per coach. In my case, I'm very close to 50% win rate overall. What means that I really get 50% win chances matches.

"Elo system is made up for single player game, you cant balance a 5v5 game with that system cuz it doesnt count the skill of a player, what champion he's gonna take, just think about an high elo player who played always same champ chose to try new one maybe he always played a ranged mage and this time he go with melee tank, k he's not anymore the skilled player he should be but the system will count him as high rating anyway. That game will be a fail."

=>And will correct his ELO slightly down. By the way, I do think that a good player trying a new character is weaker than usually but doesn't make him a total noob. Anyway, his ELO drops until it stabilize his average level between the two characters. I don't think it's a total failure.