Ive never had someone in my games more than 600 or so off
Wth you talking about Lara lol
The matchmaker does what it can with what’s available. People change their behavior and performance from match to match which can’t be taken into account.
1 Like
Even 600 is a lot. That is more than 1 rank difference.
Yeah but when youre queueing with friends 400 lower than you its bound to happen to me lol
In my solo queue roles its generally pretty close.
Its as good as it can be with reasonable wait times for QUICK play
Then why does it say in quick play to be matched against people of your skill level, when it’s evidently that it doesn’t?
I love how people are in denial that the game has seen better days.
The matchmaking got loosen once when the game started losing players.
whats your point? we all know ow1 is winding down, im playing less too. I would consider quitting if they were not changing to 5s and adding pve
The usual new heroes wouldnt have been enough
Thats impossible for everyone while keeping it quick
If they made it take longer then people would complain about that, there is no pleasing everyone
Whoa, there are competitive devs like Riot who literally communicate.
There literally is a point of comparison. It’s a matter of communication that I pointed out.
And all you can do is sit there and white-knight like a neckbeard?
I thought your kind went extinct after the sex scandal.
How does it feel to d-ride the devs?
Are you a PR slave or something?
Settle down over there
Rage and emotions arent data
Neither is your moronic take 
But the devs not communicating - oh yea, that’s a data point.
But Riot communicating? Yeah, that’s data point.
Want another?
This company managed to match a bronze against an OWL player in an arcade mode.
Yes, the video may be a bit old, but it happened.
What level of stupidity does it take for this to happen?
They are assuming independence of MMR across modes, while there is an obvious cross-correlation.
This screams either lazy or stupid to me.
This was before the scandal btw, so probably not just a skeleton crew.
It’s even simple to implement.
Use the MMR in ranked as a prior for MMR in other modes.
It’s simple Bayesian statistics.
And they don’t do it.
That’s why you see matchups like this.
And I think it still happens in QP.
The devs are communicating though
They wanted to prioritize speed over accuracy, but made such a garbage match making it drove a lot of people away from the game, further needing them to widen the already horrendous match making to make up for the lack of players caused by the bad match making.
2 Likes
Matchmaking is done based on how well each player in the game has done historically on the role that they selected. Unfortunately, there’s a lot more than that that goes into how well balanced the game will be:
- How well do the selected heroes of the team play together? If heroes are selected that don’t play well together, that puts the team at a significant disadvantage even though the players may individually be good.
- How good is the individual player on the specific hero selected? A player might be really good at Pharah but really bad at Widowmaker, but the MMR for both is shared. Maybe a player might be playing a hero they aren’t good at in order to compensate for heroes that don’t play well together, which could make the player much less good than their MMR suggests.
- How strong are the selected heroes against the enemy heroes? You could change heroes to counter what your opponents aren’t playing, but this could again force a player to play a hero they aren’t good at, again making that player perform much less well than their MMR suggests.
- How well does the team gel? Maybe all 6 players are individually very good, but they have totally incompatible play styles, which puts the team at a disadvantage.
Matchmaking is done as well as it can be, but I simply can’t see a way that it can completely prevent badly balanced situations without totally breaking what the game is meant to be.
I want to believe this - but do you think this is reasonable?
Bronze vs. OWL player in a 1v1 arcade mode where MMR should be easier to calculate because it works more like chess?
The best improvement to matchmaking they can do is improve queue times.
OK - Bronze vs GM 1v1 clearly should never happen. Does Arcade have matchmaking the way QP and Competitive do?
In QP I could see bronze players playing with GM players if they group together, in which case the game will try to compensate by putting them against oppoents that are roughly the average of the two, but I’m not going to pretend that that a GM and bronze will be well balanced against two plat players - that’s why Competitive doesn’t allow too big of a gap in SR when you queue as a group.
I will say that I am a bronze player, and while I have definitely had my share of badly matched games, I do feel like overall the matchmaking is working more or less well - Except in cases where there is an obvious smurf, I don’t feel like I’m being matched against players who are individually performing way above a bronze level.
If matchmaking were completely randomly, I’d be getting plat/diamond calibre players every game.
Do you know for sure that this is actually something that the matchmaker set up with all of the following conditions?
- Was the game was actually played in arcade mode and not an organised custom game where the bronze player actually wanted to try playing against Effect?
- Was this Effect’s normal account with full GM MMR? Does his account have lots of games played in this specific arcade mode so that the game knows just how good he is at this specific mode?
- Was this the bronze level player’s normal account with Bronze MMR? Has this player also having played lots of games in this specific arcade mode so the game knows just how bad he is at this specific mode?
I can’t read the Korean comments on the Youtube video, but I don’t see any evidence that the matchmaker actually set up this match, or even if it did, whether it was given sufficient input to make an informed decision.
Let me infer.
Pretty sure it was 1v1 mystery match.
He’s queueing for arcade mode in the video right after the match.
If this were a lobby, a lobby screen would show up with him viewing the other player’s profile.
He’s literally playing with an account named EFFECT. He was top100 material back then, so I doubt he was intentionally deranking. I bet if he did so publically, he would’ve been crucified by the community. I don’t think he’s that stupid.
At the end of the video, he shuffles through the bronze player’s profile. It seems that the player has been consistently bronze across multiple seasons.
IMO, this is inexcusable, and it happens because the devs are not hard-thresholding MMR bounds. And they’re not using bayesian priors. They’re just really stupid. Not sure how this can be disputed.