"omg how can they get so high with a 50% winrate"
"omg a 2.8k rated team with 45% win."
If you find yourself saying this, you do not understand the current rating system fully.
I explained in a post a while ago how the MMR exploiters actually do it:
Of importance for this post is that TR gains/losses are based of your TMMR & as long as your TR isn't close to your TMMR, you will not lose any TR, only TMMR/MMR.
This means you can reach high ratings as long as your TMMR is even higher, evne with a sub 50% winrate.
Now to get to the core of this "issue":
If a team of 2400TMMR starts a new team at 0TR & for the sake of the argument, their team is intended to be at 2400TMMR AND they face equal rated teams only, they SHOULD exactly have a 50% winrate till they reach 2400TR.
Why should they have more than 50%? A 2400 team facing off against other 2400 teams (aka 2 equally skilled teams) has a mathematical chance of around 50% to win (excluding random factors).
When you go head to head with an equally skilled opponent, lets say mirror matches to keep this easy, you both have a 50% of winning.
!!! Now in a zero-sum system, such as the old TBC system, this would lead to both of you sitting on the same rating. In TBC, you'd start at 1500 rating, gain 15 if you'd win, lose 15 if you'd lose. So these 2 teams in TBC would sit on 1500 forever (in a closed system, where it's just the 2 of them), even though they are worth of 2400 rating.
This isn't how the current system works though. You both start at 0, lose 0 if you lose & gain 96 if you win, until your TR comes close enough to your TMMR, then the gains start decreasing slightly & you actually start to lose rating. Once these 2400 teams reach 2400 & they still go head to head, they will maintain a 50% winrate @2400.
As an illustration for those who know, this is exactly what the russian 3600 teams did. They grinded their MMR up skyhigh, then split even against eachother. Since they keep on gaining TR until they come close enough to their TMMR, they could easily reach 3600 rating by splitting even (after having abused their MMR up ofcourse).
The only teams that, in this very simplified theoritical example, should have more than 50% winrate are teams that either
1) Are below their TMMR indicates (example: http://eu.battle.net...NNYANNYANNYANN/ )
2) Face only lower rated teams
3) Are significantly above the rest of the ladder (ex, a 3600 worthy team on a ladder where except for them the highest team is 3000), mostly rank 1 contenders (example: http://eu.battle.net.../3v3/DNAW trio/ ).
Even if a team has a negative win/loss to their rating (example:http://eu.battle.net/wow/en/arena/kazzak/3v3/Pala%20Drunk%20GG/ ), it doesn't say anything about how good or bad their team actually is.
It can mean several things:
1) The average TMMR they faced upto grinding that rating was higher than their own TMMR.
2) Their TMMR was slightly higher than what they were worth off & because of that, they lost a few games more than they won until they established their actual TR.
So why do people have such issues with these winrates then?
First of all, this is because in every (e-)sport, teams that are considered good always have astonishing winratios. It's just what's common to people & what is considered "normal".
Secondly, the TBC MMR system was all about winratios. You could get high without a huge winratio. Top 20 teams would almost always have atleast 75%+ winrates. People still have this state of mind where they believe winratios matter for everybody & that you're worthless with something such as a 50% winrate.
Like I tried to explain, this isn't true at all, a team with a 45% winrate @ 2900TR, 2900TMMR can be amazingly good, yet they faced teams that on average had 3000TMMR.
Edited by Charred, 16 July 2011 - 12:02 PM.