I think I figured it out, but for one, there is no LOSERS QUEUE… I think most of the words “we” use are just due lack of details of whats actually happening and more of a “buzz word” so people may understand quickly the feeling of something but not actually have that “thing” actually being implemented.
lol if any blizzard shills or rigged person mentions an actual losers queue after that explanation in this thread… I may not respond.
Ok… so aside from “Role Avg MMR” grouping… I think there’s another kind of “grouping” or lack there of…
A little 101: MMR is stored as a variable AND a sigma range / skill bell curve (internal to everyone, not the WHOLE skill bell curve).
I wonder… what if… stomps happen if 1 team is at their “lower” skill curve, while another team is at the higher end of their personal skill curve.
OR… lets say your at your high end, do you attract those at their lower end?
OR is it just random, and thats why stomps dont happen ALL the time?
The question is, what would be the optimal team make up WITH and Against? Fair would be the whole lobby is either at the top of their curve or at the bottom…
I ask cause now that Role avg is going to be a thing, but if one team is at the top and the other team is at the bottom… it might not make a difference.
Hey my friend,
do you understand how a bell curve works?
Yes… but you know I am not talking about the population bell curve… Each individual has their own skill bell curve.
The fact that no1 has responded this means exactly three things…
- I probably chose the worst word combination to describe my idea (highly likely).
- No1 wants to answer because no1 understands (probably cause I am not a great words smith).
- Like Role Avg… this maybe an overlooked cause of stomps and no1 wants to admit it.
I kid I kid… However you have to understand that there are reasons why someone maybe at their lower end of their personal spectrum… and it might not be because of THEIR skill, it might be because of other factors. Like Balance… like meta… like overall bad comps, leavers, throwers, rank integrity, lag, yadda yadda… So to match teams on purpose that may have all players on 1 team in their lower personal quadrant vs a team whose all in their upper quadrants maybe the cause of “loss streaks”, stomps… the feeling that a “switch” flicks on after you reach your personal best to set you up for a loss streak.
Imagine you surpass your upper quadrant and the game now positions you in a higher bell curve… but puts you in a “new” low quadrant, however everyone on your team is in their low quadrant. I just find it a little “unfair” to have a lower quadrant team vs a higher quadrant team.
Tf is a skill bellcurve? You have good days and Bad days where your performance might flactuate. That has nothing to do with a bell curve. Your mmr sets a range for the matchmaker to fill Out your lobby, which can lead to situations, where you get boosted players, smurfs and groups in your lobby. Stomps happen in every competitive environment. I’ts not a hidden rigged system. The reason for that is that on or multiple people perform badly on one side and the other team Just outperforms them
MMR is stored as a value and a uncertainty. Everyone has a hidden MMR and a Hidden “bell curve”.
The uncertainty is just a value.
As you play more games that value shrinks over time.
(http://ratingupdate.info/player/2EE13E63DF86C8D/PO)
(this is just the person whos number 1 on the leaderboard, i havent played since December so I cant pull my data)
This is a fan made site that calculates and displays MMR for Guilty Gear Strive.
When you first start playing your uncertaintly is in the 100s.
As you play more games (around 20) it drops to under a hundred and after like 50 games it levels out to around +/-20
Think of plot points on a scatter graph with those little error bars on each point.
Loser Queue is a “feeling” imo.
What could explain that feeling :
- MMR is a value tied to played and its relative as the devs said. That means the Match Maker doesn’t think as “your MMR is 500, so I’ll pick people with similar MMR”. It seeks players as “I got you one mate who has +50 MMR” ; look at average ranks for both teams and try to get even average numbers.
The bell graph devs showed us is showing that Matchmaking chooses people with a lower MMR to be put in your team (between -1.0 and -0.5 ; note that we don’t know if its a multiplier or an added value).
Putting you with higher MMR mates can be thought as helpful or harmful. Better mates would carry you, but better mates can make you feel useless and allow you to make big mistakes and still win. Not mentioning also, taht if the both team average ranks are even, that means your opponents will be stronger than you too.
- As the devs mentionned they try to use predictability in the matchmaking system, that means the matchmaking will decide to put you with higher MMR mates or lower or both.
Afaik, we don’t know how the MMR decides it.
So what’s happens with matchmaking : devs showed usually it picks lower MMR players than us. Not much lower but still lower.
The matchmaking has a “prediction” system to avoid winstreak and losestreak. Yet devs admitted that predictability is harder when the match rate gets lower. The worse players are, the harder it gets to predict who would win.
So when the prediction fails, like matchmaking goes with “okay you need a win” but fails, you as a player feel desperate and think about a Loser Queue. But according to devs, its more a mistake, an issue of the predictability of the games than an intended Loser Queue.
- There’s also a psychological explanation. The more you lose, the more tilted you can be, so the worse you’ll play.
How could they fix that ?
- Imo, they can show both MMR and skill rating. Problem with showing just Skill Rating could lead to player being over-cautious for the sae of their stats. If they show that players need to win, rather than having good stats, maybe some players would accept to swap heroes, to try different things.
They could rework the whole matchmaking system. Is winrate the only variable the game should look at to give a feeling of balanced/fun games ?
The problem with winrate exclusivity is that, you wouldn’t be encouraged to learn a new hero. I.e. I’m main Moira, she has the best winrate for me. I was willing to try to learn Baptiste. Ofc, as I’m a bad player, I lost a lot with him. So my MMR lowered. So now I’m in that case : if I pick Moira, I can destroy some games. If I pick someone else, my stats are very bad and I don’t feel good. I can tell it even more if I play half of the game with any hero then I pick Moira for the 2nd half.
And it works for your mates too : you’ve lost your last games. You get matched with “decent” players but, tough luck, they decide to not play their main heroes… You might lose again…
-
One way would be to lock heroes before the game starting… But I really don’t like that idea and it removes the ability of swapping heroes which is an essential part of OW.
-
They could find an other way to define each player performances. Fire system is imo a possibility for that. And that’s what I wish for. I feel like fire system has been underrated by devs since OW1 day1 but its a good way to tell who’s doing good without PotG, without the stats.
correct, but your not taking it to the next abstract level.
You have a number line of 1-10, you MMR is 5, with an uncertainty of ±3. Your bell curve spans from 2-8, where 5 is your peak… We’re not saying different things.
Lets say you lost a game. Now your uncertainty goes to ± 2, but MMR goes to 4… and so and so on…
Albeit the the hardest part is the “moving” window after every game. In the example above, 5 going to 4, that person is near their previous lower quadrant window.
This goes back to NOT having that uncertainty close up so fast. Or have conditions that allow it to be flexible.
It put me on a losing streak and I lost 3 ranks. My performance has been good personally. I keep getting subpar teams. I had just climbed back to get my best rank too. This is the second time it’s happened this season. I don’t get what I can do to fix this as support.