Is -20 supposed to happen after 200 games+ played and less then 3 min q pop timer

[Abelisk.5148]

Yeah i don't get why teammates don't affect rating loss only the enemy team. Because you can have some pretty shit people on your team that you just cannot possibly save (especially when solo queuing) that cause your rating to tank when losing.

[Sailorz.5426]

@Trevor Boyer.6524 said:

@JusticeRetroHunter.7684 said:

@"voltaicbore.8012" said:I could be reading the matchmaking algorithm on the wiki entirely wrong, but I think it actually allows for teams that are rated up to 1200 points apart to end up in a match against each other. I am not sure if that "roster rating" variable is on the same scale as our personal MMRs, but that seems to be the case. That's... that's an absolutely enormous score difference. So that might explain why the matchmaker is so quick to pop poorly matched games for you - it just doesn't take that long to patch together a team that's within 1200 points of yours.

This is mostly (educated) guesswork on my part from reading

the wiki post on the matchmaker

. If someone knows I'm wrong about this, I'd be interested in hearing the explanation.Something real fishy is going on, and i can't quit put my hand on it...but my guess is that there is a statistical reason for why we are receiving higher yielding losses in these lower rating ranges.

You've got that right, and I'll be reeeeal careful what I say here, but something seems.. unfair vs. certain accounts. Whether it's an accident like some kind of bug or just old outdated functions within the algorithm like "how invisible MMR effects your match making under rating and how total matches played effects MMR directly", something is wrong with it.

Now you guys will probably all take this the wrong way when I say it, but I am being serious so hear me out on it. If you were to watch my stream and follow the AT footage, you'd see that I consistently win 1v1s on side nodes vs. some of the best players in NA, regardless of class vs. class counters. There are legit only about 10 guys in NA regardless of class played, that I even feel threatened by when approaching a 1v1 situation. But yet somehow, these other players will have 70% win rates and be like 1650-1700+ rating, whereas when I had 70% win rate this season at around 60 games played last week at some point, I was only 1525 rating. Somehow these other players at 70% win rates are riding p2-3 but my 70% rates I am riding constantly top g3 bottom p1. I mean.... what? Please explain this to me. Because a lot of those players actually do go out of their way to queue dodge other good players and they only play against lower players so they can farm. Their gains should be smaller than mine and their losses heavier. So how in the hell are they achieving 1700 ratings with 70% win rate vs. low rated kids but my 70% win rate vs. NA prime time platinum players yields a 1525 rating? What? Please explain the secret of what's going on here. <- I am being serious, what the hell is that?

Tbh I should stream a session of me playing ranked to show the match quality and gains/losses that I get on this old year 1 day 1 created account with 20,000 games played. I'm telling you something isn't right with it. As much is visible when I go on an alt account and easily play into p2 margins with less than half the kind of monkey business going on that I see on my main.

I 100% know how you feel. The shameless plug won't help your case though. Followed C: .

[Eddbopkins.2630]

yeah i got a -19 the other day after getting -10 to like -16 but all of a sudden after a hard fought game and i mean hard fought were i lost by 5 points i lose -19. how does the matchmaker calculate points give and points lost after each game. i feel its time to rework the match maker to have better teams. no more expanding que. i dont want people who are less me on my team and i feel this happens way to often. we need to be able to see everyone's mmr by now. its been long enough. Whatever mmr the system uses to make a team those are the numbers we all should see. HONESTY IS TRANSPARENCY!

also i play against a guy and lose to him 3 times, why does the match maker put me against the same person a 4th time? they want me, to wish the matchmaking gods deed?i havnt won a game in 10+ matches. i feel i keep getting people who are just worse and worse on my team, and i keep playing the same people. i want transparency with the match maker its been long enough with the bs.

[Luthan.5236]

From the discussions in the past I know that it compares "own rating" vs "average/mean reating of the enemy team". I usually get 12 or 13. But I also have seen 15 ... usually for the losses (with - only and rarely with +).

This might be if weaker guys queued into the enemy team (maybe as duo queue even) and the matchmaker could not make evenly matched teams. Or if you are above average in your own team. (And the enemy team has a simliar rating on average.) Really annoying when you get roflstomped and thing "oh against some stronger enemy I will lose less" and then it punishes you by actually deducting more. :D

Yeah ... the system is flawed here. At least in terms of motivating players. (It is demotivating. Should at least be hidden and not shown.)

In this example though it seems there was a close fight: Probably the enemy was not that much stronger. So easier to understand why the rate loss might be higher. (Though here it is annoying that it doesn't at least recognize that you fought hard and "almost won".)

Edit: Low queue time could just mean the others were long in the queue and you were one of the last players that got added to finally have 10. + there is always the possibility that very strong player is rated low. He will make your team lose. While making you lose a lot of rating cause you were supposed to actually win against him because the system didn't know (yet) that he actually is strong. That is the problem. That is why I would suggest to hide ratings so it won't demotivate. Too many players are complaining about it. And flaming towards the own team happens most of the time causs someone takes the stuff "too serious" (cause he cares about his rating).

[Mega.5791]

@"voltaicbore.8012" said:I could be reading the matchmaking algorithm on the wiki entirely wrong, but I think it actually allows for teams that are rated up to 1200 points apart to end up in a match against each other. I am not sure if that "roster rating" variable is on the same scale as our personal MMRs, but that seems to be the case. That's... that's an absolutely enormous score difference. So that might explain why the matchmaker is so quick to pop poorly matched games for you - it just doesn't take that long to patch together a team that's within 1200 points of yours.

This is mostly (educated) guesswork on my part from reading the wiki post on the matchmaker. If someone knows I'm wrong about this, I'd be interested in hearing the explanation.

Indeed. However, you missed one more detail:

"Padding is added every second you wait in the queue after Filter/Rating/@start has passed."

The padding starts only after 5 minutes. However, I cannot find a "standard maximum rating difference between rosters". Only that it supposedly starts at 25 - which seems unreasonable. After 5 minutes, games start to get pretty weird. -20 is far from uncommon, and regarding the low population of US servers, I don't know why people are surprised.

However: GLICKO can only work with what it has. The matchmaker struggles with some of the settings like switching classes, duoQ and all that. Maybe Anet should reduce the "standard maximum rating difference between rosters" to a very low value for an off season and see how players like getting no games for... a looong time. This will mainly affect the top and bottom 10%.

At this point: Hide the rating changes and only show the final rating at the end of seasons. Maybe people will get triggered less.

Or maybe they will rage for a week after the season has ended because "what I am much better than THAT GUY the system is rigged!" conspiracy theories.

[Master Ketsu.4569]

If you look at the leaderboards you will notice that there are still people in Gold 3 in the top 250. Within only 10 days of the season ending, this is normally not the case. The fact is Population in ranked is low enough that once you get up to around Plat-II you find yourself in an uphill slope as every win is +7~10 and every loss is -16~20. I'm currently in this bracket myself,  it is what it is.

 

When pop is low you generally are going to have lopsided games and most people will find themselves ranked lower than usual. As it turns out, the current "Zerg+Res exploit" meta just isn't very fun.

Edited April 30, 2021 by Master Ketsu.4569

[Robban.1256]

On 4/19/2021 at 9:15 PM, Lighter.5631 said:

why is -20 happening that's not shortly after placement.my rating isnt even that high, is this game dead and why is this not capped

my team was full of bots and enemy players at least know how to press skills..am i supposed to be Q'ed into this i supposed gold 1 game with less then 3 min Q pop time.i think the system is broken.i think there supposed to be rating range cap

The Glicko system isn't just based on your current rating, it also takes into account the volatility of that rating based on the standard deviation of your results over time.The variance in point loss/gain for similar point differences has to do with each individual players' rating deviation (uncertainty). Players with high RD will earn or lose more points when they win or lose respectively.e.g. -If you loose against a weaker opponent, or win against a stronger opponent, that will affect your rating more than loosing to a stronger opponent or beating a weaker opponent.

I will not reprint the Glicko equations here because they are much more complex than the Elo equation, but for the mathematically curious an overview that includes the equations can be found here http://www.glicko.net/glicko.html

There's a graduate student master's thesis on the topic if you're interested.

http://rhetoricstudios.com/downloads/AbstractingGlicko2ForTeamGames.pdf

 

Glicko is mathematically better than Elo, but requires more calculation. Elo was conceived around 1970 and could be calculated without a computer. Glicko is only practical with a computer.Glicko and Elo should reach the same rating, but Glicko reaches the right rating faster. Glicko-2 is more modern system than Elo and encounters such factors as new people in the pool and changes in strength to adjust speed of rating change.

 

If two continents implement Glicko or Elo and the players of the two continents never play one another, then when they start playing intercontinental it will always show that one continent is overrated and one is underrated. Only if they play intercontinental will the ratings on both continents equalize.That is also the case with FIDE rating and USCF rating: USCF is overrated as compared to FIDE.

 

''@Ben Phongluangtham.1065 said:In a recent random sample of 100,000 matches, we found that in approximately 95% of matches, the difference between the average skill rating of each team was less than 50 points. The matchmaker is doing a good job in most cases. Things get more problematic at the very low and very high skill ratings. Our change to duo queue for 1600+ ranked players is part of our efforts to address this. In addition to that modification, we’re working on some fine tuning on the matchmaker. Our simulation with the proposed changes extended the favorable difference ratio mentioned above from 95% to over 99% of matches. I can’t give you specific dates on when these changes go live, but we’ll be looking to trial them on the unranked queue somewhat soon™.

One thing to keep in mind is that just because the average skill rating of each team is close, that doesn’t mean you won’t have a blowout match. Some maps just tend to snowball, some players tend to give up when they get a bit behind, etc. This can lead to a blowout even if the average skill rating of each team was fairly close.

I know we have top stats, but they don't generally tell the whole story. Individual performance is a very situational thing. If we tried to adjust ratings based on stats, we'd have people chasing stats rather than the win. That's why we value the win over everything else.

@Ben Phongluangtham.1065 said:@Forsty.7968 is correct in that we use the composite opponent method for rating changes. I think we may have tried the composite team method in the past and it didn't work well. I could be wrong about that, as it was before I joined the PvP team. The problem with composite teams is that it basically never gets people to a rating that reflects their real skill.''

And for you who doesn't know who Ben is he is one of the pvp devs.

Edited April 30, 2021 by Robban.1256

[Robban.1256]

On 4/21/2021 at 6:26 AM, Abelisk.5148 said:

Yeah i don't get why teammates don't affect rating loss only the enemy team. Because you can have some pretty kitten people on your team that you just cannot possibly save (especially when solo queuing) that cause your rating to tank when losing.

see my note above about team vs team

[Trevor Boyer.6524]

Oh yeah, to follow up with this:

 

It's too bad because I accidentally deleted this video in my stream, but during the forum downtime I had caught on footage a match in ranked where I was currently 1556 rating and I was against a team that had the currently rank 4 player on it, who was like 1700+ at the time and his team was stacked with all p2+ aside from one player who must have been gold 3 or bottom plat. My team was full of random PUG aside from one other person whom I know to play between plat1-2. When I lost the match I received -10 points. Then when I won the next match against players roughly the same rating as my 1546, I received +10 points. I lost the next match at again 1556 but this time against players of my same rating, and received -10. 

 

Soooo... explain to me what happened there? Because in every previous season I remember losing only -4 or -5 when 1700+ beats me if I am 1550ish. It's too bad I had deleted that video, I was specifically saving it for this forum to reopen. Oh well. I'll probably catch something like it again. Anyway so yeah, something doesn't feel right with these gains & losses this season.

Edited May 1, 2021 by Trevor Boyer.6524