Who said those sites where 100% currect ?
Those sites relay on people uploading thier games. The real data Blizzard keep for themself.
Tassadar and nazeebo are both a mage.
If winrates are considered, they could have applied much better matchmaking algoritms where equal winrates are more likely to get matched together (which should be relatively easy to apply). So i think it instead is looking at just the MMR changes. Where even 35% winrate might only translate to something like -50. Which is a relatively small adjustment here. Especialy because with such low MMR, a loss isnt counted very hard in those matches.
The algoritm could even have gotten a secondary check afterward to potentialy still shift a few players around. So regardless of them being matched in that match, they might shuffle an equal role to the other side to balance the average MMR. This is at most 25 checks if done properly and faces its worst case scenario (5x5 checks)! In ideal cases it only has to check 1 player and 1 enemy player (and when both arent in a stack) which both have the same role, switch them around and look at the resulting average MMR. If that MMR gap becomes smaller, it can decide the switch was good. It could even only limit itself to just 1 change and take the highest effect of the potential 5 switches it gets.
These types of match ups just shows that no corrections are taking place and it just relies on randomness. And sure, this can go both in a positive or negative way (if the highest MMR is a solo, the solo team has a higher chance of getting that one aswel - which we do see as nazeebo is the 4th place of the best solo player according to MMR).
But the problem is that usualy there is only 1 extreme, and the 5 solo’s in this case have a 5/6 chance to get that extreme one on their side. And especialy with bad players. Its more noticed. The stack generaly already exceed the solo because of being grouped.
This is an example where the matchmaker simply didnt do its job of providing a balanced game. Not just unlucky, or QM, but i would consider it just bad coding or lazyness.
A smurf at a sub 40% winrate is not an advantage in almost any case. It shows a trend of dropping in MMR. And at hero level 37 its very clear that there simply arent enough matches to truly have a balanced MMR.
1 Like
it is reliable to certain extend, and it is reliable the more a person upload their games.
and i have uploaded my most recent 3000+ games out of my almost 6000 games throughout 6 years.
aram and vs ai seems very popular on eu usually10-20s queue time
though yeah some dcer here too which i believe is a problem of ppl playing on the wrong server and the merge of eu east and west was just madness
1 Like
The game allows for a player to have about half the mmr of the highest mmr player in the team. This is due to low population forcing them to just slap any match together.
Also an actual smurf isnt going to have 30+ % winrate, its the otherway around so in this case they likely were just low mmr players.
1 Like
yea, kinda know this is a norm, but still dont understand how is both team’s player skill rating is not more balanced. like, i understand if they enemy team is 5 stacks, but they are 4 with an experienced average skill player, while my team has both newbie with low winrate.
thanks haha, had a debate up there and some people are convinced it is very possible. created an alt which im not proud of to test out this theory. after around 50 games, with 60% winrate, i get into my main’s skill rating with enemy stacks that i was very familiar with.