Why are these even a thing? Why purposely pair players against each other when even the system knows the matchup is imbalanced?
15 Likes
I donât know why people think we have this perfect world with the perfect number of players to always create perfect matches.
Guess what if my team wins the âexpectedâ for you (uphill battle) for me Ill gladly take that bonus SR from your hands.
The matchmaker put me against you because it thought I was weaker then you but still wanted to try and create a âBalancedâ game you donât want the better player staying in lower ranks ruining even more peoples games now do you?
3 Likes
Your comment is an oxymoron. It never tried to make a balanced game, because it purposely paired players it acknowledged were not evenly skilled against each other, with the expectation of who was going to win.
Thereâs enough players going in game that this should basically never be a thing.
4 Likes
NO NO THERE IS NOT! WHERE ARE THESE imaginary players coming from!
And our ranks are constantly shifting and moving.
Thereâs always slight discrepancys here and there sometimes even MAJOR ones, like there main having balanced changes, them improving at the game etc etc.
2 Likes
I am not sure why theyâre a thing either, because I get âexpectedâ on games that I had to fight for my life to win, and âreversalâ on games where my team can do nothing but get curbstomped on the spawn doorsteps over and over.
5 Likes
Because unless you want to wait an EXTREMELY long time, matchmaker is going to keep people in a range instead of a 1:1 ratio to 5 decimal places.
5 people, 3 roles, many regions.
The funny part is it seems like only recently theyâve acknowledged a silver DPS is not worth a Silver tank.
But in any matchmaker Iâd bet someone who writes them would tell you that even the number or value it assigns you isnât perfect. Itâs still an approximation and treated as such.
3 Likes
Because it is a machine, not a magic genie the numbers in your profiles showed that that game was supposed to be closeâŚ
Thats it⌠thats all it has.
It dosenât know what heros work together, it dosenât know the meta.
It knows beep boop beep beep boop.
3 Likes
It might not work like this, but a simple non-rigged variant would look something like this:
Randomly draw 10 players from within a narrow rank range, then randomly assign them into two teams without regards to detailed mmr balance.
So now you have a team A mmr avg, and a team B mmr avg.
Say, 2450 and 2520.
The game would have not yet used any knowledge of match outcome to ârigâ the match. i.e., It hasnât rigged people around or handicapped slots directly to force a 50-50 lobby with tight mmr.
Yet based on the small gap between randomly assigned teams (as well as your own mmr, and team avg mmr), you could assign more/less mmr adjustment based on the outcome after it finally happens.
In this example, players with mmr lower then the team average, lower than lobby average, etcâŚare expected to be underdogs and might gain more per win while someone evaluated much higher would lose more for a loss.
It sounds like something that could/should work in principle, and even work in non-rigged free open-source ladders. Seeing lots of expectation reversals would be a sign of poor sampling i.e. lowpop or ladder corruption (smurfs). The frequency of which imply a host of other ladder problems.
2 Likes
Thatâs 5 games in a row now where it said âexpectedâ or âconsolationâ. Literally 5 games in a row were decided not by the players, but by the matchmaker purposely pairing bad players against good players.
3 games consolation, 2 games expected, so far.
1 Like
Did you stomp the âexpectedâ games and almost win the âreversalâ games? Because my experience with these modifiers is that they are meaningless. The games donât play out any differently than they would have without the modifiers.
As Iâve said several times, and Iâll shout it from the rooftops - the âmodifiersâ are essentially meaningless buzzwords adding to the smoke and mirrors surrounding the system, to give you âinsightâ as to why you gained or lost as much as you did - but the gains and losses are still all over the map whether there are modifiers shown or not.
2 Likes
How âmuchâ expected or reversal is the question. The odds of getting a match with balanced teams is small, compared to getting matches with a small mmr gap one way or the other, assuming random sampling.
But if the frequency AND amplitude of these are high. Say, 3/4 games are large expected or large reversalâŚthen thatâs a symptom of hidden problems. Things like lowpop dead game, lobbies ft. mmr that is nowhere near the SR (i.e. alts/smurfs), bad algs/code, rigging, etc.
1 Like
Sorry, I typoâd. I meant the last 5 games were âExpectedâ or âConsolationâ. Edited and fixed it.
Also, now Iâm up to 6 games in a row this session.
3 expected, 3 consolation.
And yes, every game has been a pretty clear stomp so far. No game has gotten to a third round. One was kind of close, but they were still 30m off from catching up on Push. But you could count that as not a stomp, if you want. But thatâs still 5/6.
oof
IDK i get stomps all the time too and half the time they have random modifiers that donât even make sense, so my experiences has been that they just donât seem to matter.
Get rolled by the enemy team but was âexpectedâ to win. Sure thing Mr. Matchmaker.
7th game in a row now. Consolation. We barely won the first point, but then their healer swapped off of Mercy and we then lost all 3 followup points.
7 games now, this play session alone, that has been determined before the doors even opened up, and Blizzard acknowledges it.
Most of my games are âExpectedâ. I rarely ever get uphill battle. If I lose once, I get reversal, and if I lose again after that due to poor matchmaking, I get Volatile.
The modifiers suck, and shouldnât exist. You should gain a set amount for winning and losing, and get a slight boost if youâre on a winstreak.
4 Likes
The modifiers have been in the game since ow1 just not visable.
Also no a set amount would be boring, I donât wanna go against the team thats pretending to be the OWL win and not get bonus rewarded for it thatâs a waste of my time and everyone elses.
9 games in a row were not balanced match ups. 5 consolations, 3 expected, and 1 reversal, which would have been an expected but we had a hard thrower so nothing could be done.
Slightly unbalanced is part of the competition. Natural odds are byproducts of fair matchmaking without rigging and handicapping to force 50-50. Your skill update is supposed to represent your signal over the background. Randomly sampled lobbies of 10 players randomed onto 2 teams gives some mmr mismatch which you get modifiers on based on how much above/below the lobby mmr you were going in.
You have the frequency but need to know the % offsets (amplitudes). If I roll the dice a bunch of times, shipping people from 2400-2600 mmr onto 2 teams, Iâm going to see small offsets like +/- 150mmr for team averages. Most games will be not finely balanced. For small SR range thatâs only like 150mmr offsets. In and of itself, itâs not supposed to be an appreciable difference in lobby playout. So the expected/reversed is just slight modifier to your +/- based on what you came in with and what the random averages turned out to be.
Itâs a good intention drowned out in the swamp of everyting else including buggy code we canât know, audit, or help with.
1 Like
Imo it shouldnât be a thing. If the matchmaker decided this is a fair match and it canât predict the outcome, the rating should not be adjusted assuming a specific outcome.
This would also take into account that the rating is probably not perfectly accurate.
Imo only uncertainty (magnitude of any adjustment) and streaks (velocity) make some sense.
Well I donât particularly enjoy being punished for not being able to carry a Tank that is clearly new to the role.
1 Like