Talk:Elo rating system/@comment-188.178.187.14-20120126081725/@comment-3014924-20120126090549

About post 1: You are wrong at this sentence:

"Using rating system from Chess is the biggest mistake RIOT has made"

You are right about that random position within a categorie (tough deciding there are 2 is false either) because, yes, your personal skill affects way less your win probability than it chess.

But where are you mistaken? You're mistaken about the fact you could come up with a better system than ELO.

And now part 2: You could adjust and get the ELO righter by affining the probability of win/loss:

Like you said, in chess, there's no random factor, so a guy twice as strong as person 1 has a probability to win of p=D/(1+D) (D=2 if player A is twice as strong).

Usually, in LOL, matchups will be of same ELO, meaning win chances=50% (because they use the chess fomula)

Now they could improve this a bit: -Make statistical analysis about win chances depeding on the team composition -Give every player an hidden "role" ELO (like, you're 1000 with AD carry, 800 as tank...) -Modify according to the number of games played with the picked character -Are you duo-queuing? and so on...

It would change a bit the ELO and well, slightly change the discrepency around your real ELO value.

Now, I don't bet they would do this. Too much work, has to be updated.

And if they don't, ELO remains best system around.

And if you really understand maths and probabilities, you would get that.