High Rating Disparity Duo's Cause Unbalanced Matches, Limit the Gap to 100 or Less Rating Points

[lightstalker.1498]

It's not friends playing together, it's a person who wants to climb with a buddy on a low rated alt to get easier matches. 1600-1700 or higher rated player queues with a false silver, who is currently rated in the 1000s or 1100s. You know what kind of match that gets them. Thing is that same low rated account loses alot of matches, but not when they duo with the high rated buddy.

 

Suggestion is to lock up duos to 100 rating points or less disparity. Matches will be more fair, and you will see less wacky stuff in games. I don't want duos to go away, but players need to queue with others of the same caliber and not try to manipulate for easier matches, you know, I know, we all know this is cheating even if  (nerd voice) "technically" it's allowed. Time to make it "not allowed".

 

Besides if only so many people duo, then you will see less bullcrap duo's in your games, and have the odd silver rated holo or herald named FHkjshdy, jabbarooing 3 of your team mates while you are wondering what the hell is going on. Limit the gap to 100 rating points or less.

 

Easy fix.

[Avatar.3568]

It's a cool idea

[Fueki.4753]

In my opinion, limiting who you can queue with isn't a good solution.

Having both players queue at the higher rating of the two might be a better solution. For example, if one player with a rating of 1700 and one player with a rating of 900 queue together, the MMR will treat them as if both players had a rating of 1700.

Unlike your idea, this allows people to play together independently from their personal ratings.

Edited December 30, 2021 by Fueki.4753

[Stand The Wall.6987]

makes sense.

[covahlam.6391]

What about a conditional threshold where the higher rated player's rating is used for both players if the gap between partners is greater than say 200, otherwise the duo's ratings are treated as they are today? I mean, there will always be issues around the margin with any threshold used and there is the fact that rating is a nonlinear representation of skill. Still, I think something like the above would help alleviate manipulation from high-low duos without ruining things for the majority of players.

[Kuma.1503]

That's what LoL does. It doesn't let a diamond player queue up for ranked with a gold because it would cause the exact same problems you mentioned here. Good call.

Edited December 30, 2021 by Kuma.1503

[aelska.4609]

Game is not balanced around competitive pvp. Annoying the entire playerbase to punish the very top few percent is not a good idea, especially in a MMO. 

I was the first one to support the idea of reducing ranked queue to 2v2 but I had the time to realize it was a bad idea. Winning against premades is a satisfaction that already disappeared from the game, as well as playing with a hand of friends.

[Crinn.7864]

Ranked PvP uses the highest rated player's rating. It does not average. Unranked however does use an average.

[lightstalker.1498]

43 minutes ago, Crinn.7864 said:

Ranked PvP uses the highest rated player's rating. It does not average. Unranked however does use an average.

 

Nah. This is not true. It looks at each person individually, and builds a team average, regardless if they are in a duo or not. It doesn't even try to put one duo on each team, that is why you have some games where one team has two duos and the other has none.

 

That should be proof enough, but in case it isn't, here is the configuration.

 

<Arena name="Ranked Arena">
  <Queue>
    <RosterSize min="1" max="2"/>
    <Iteration interval="30s" rosters="100" limit="250ms"/>
    <Potentials min="20" max="500" falloff="0.375" start="1m" end="3m"/>
    <Rating start="5m" end="10m" max="1200" min="25"/>
    <Power curve="1" percent="1"/>
    <Rank min="20"/>
  </Queue>
  <Matcher type="Team">
    <Age seconds="2"/>
    <RosterSize max-diff="3" distance="-100" perfect-fit="0"/>
    <Rank distance="0"/>
    <Rating distance="-10"/>
    <Profession max="2" common="-100" unique="0" matching="100"/>
    <Dishonor distance="-100" stack="-50"/>
    <GuildTeam affinity="50"/>
    <Games max="500" distance="-0.25"/>
  </Matcher>
</Arena>

 

Notice how is is looking at the individual, regardless of their queueu size, it does this for each player so it can try to build a fair match of players close to the same caliber. If I play at the plat level and I queue up with a guildie who is at the lowest silver tier, I definetely can see a difference in the play.

[Crinn.7864]

37 minutes ago, lightstalker.1498 said:

 

Nah. This is not true. It looks at each person individually, and builds a team average, regardless if they are in a duo or not. It doesn't even try to put one duo on each team, that is why you have some games where one team has two duos and the other has none.

 

That should be proof enough, but in case it isn't, here is the configuration.

 

Notice how is is looking at the individual, regardless of their queueu size, it does this for each player so it can try to build a fair match of players close to the same caliber. If I play at the plat level and I queue up with a guildie who is at the lowest silver tier, I definetely can see a difference in the play.

 

 

For our purposes the config isn't what you want to be looking at. You want to be looking at the Pseudo-Code which is the next section on the wiki page. The relevant function being the scoreRoster function at the bottom. 

 

def scoreRoster(roster, team, maxRosterSize, config):
  score = 0
  
  # adjust score by time queued
  score += roster.age * config.age.seconds
  
  # adjust score by rating difference
  distance = abs(team.averageRating - roster.rating)
  score += distance * config.rating.distance
  
  # adjust score by games played difference
  distance = abs(team.averageGames - roster.games)
  score += distance * config.rating.distance
  
  # adjust score by rank difference
  distance = abs(team.averageRank - roster.rank)
  score += distance * config.rank.distance
    
  # adjust score by roster size difference
  distance = abs(maxRosterSize - len(roster))
  score += distance * config.rosterSize.distance
  
  # adjust score by profession counts
  for profession in allProfessions:
    # roster has none of these professions
    if roster.count(profession) == 0:
      continue
    
    # too many of the same profession
    totalCount = roster.count(profession) + team.count(profession)
    if totalCount > config.professions.max:
      score += (totalCount - config.professions.max) * config.professions.common
      
    # otherwise favor the variety
    elif team.count(profession) == 0:
      score += config.professions.unique
      
  return score

Unfortunately, even this doesn't tell us what we want to know because we need to know how roster.rank and roster.rating are implemented. No other information is given. There is the XPath section which has definitions for Scoring/Rating/@distance and Scoring/Rank/@distance. I have no experience with XPath and I don't know if it is referring to the same thing and the pseudo-code or not. If it is, then your position is correct. However I have a very distinct memory of ArenaNet patching the ranked matchmaker to only use the highest player in the group due to people exploiting the hell out of it. IIRC this was sometime in the early seasons.

[Multicolorhipster.9751]

Its a good idea. Should have been like this from the start.

Ranked in gw2 was eventually structured like Ranked in pretty much every other game anyway, so why not treat teams as any other game would?

 

Still, 100 rating is a little strict. The matchmaker can go well-beyond a 100 rating difference just SoloQing.

And you'd still have a few power couples out there ruining the game for pretty much everyone. 

 

That being said, split queues would be perfect for this game imo. IE SoloQ and Teams as separate ranked arenas for all Ranked modes(5v5 conquest, 3v3/2v2).

 

Then there wouldn't be any need to restrict DuoQ over and over again.

People could play conquest with more than just 1 other person.

Teams could have their own ranked ladder, leaderboard, titles, etc.

Nobody would play with an inherent advantage over anyone else in their match.

 

Think it over 👍

[Crab Fear.1624]

5 minutes ago, Multicolorhipster.9751 said:

Then there wouldn't be any need to restrict DuoQ over and over again.

People could play conquest with more than just 1 other person.

Teams could have their own ranked ladder, leaderboard, titles, etc.

Nobody would play with an inherent advantage over anyone else in their match.

 

Is this an idea to separate the queues?

So would the option be Solo only and 1-5 queue?

I'm a bit slow, so trying to imagine how it could make a team of five when some groups are two, and some would be four.

I guess some solo players would still hop into the 1-5, but matches probably would be slower for a while.

Either way, if thy made one choice to be solo only, and the other 1-5, it would work.

But, both would be longer queues for a while, until people realized that Anet had divided the queues, and it might bring back some people.

Also, it could show what people did more.

Yeah, for quality teams, I would wait longer times to get matches.

Even if getting rekt, i think it could revive the guild scene.

 

 

Only thing to resolve is the leaderboard / season rewards.

 

 

[Multicolorhipster.9751]

2 hours ago, Crab Fear.1624 said:

 

Is this an idea to separate the queues?

So would the option be Solo only and 1-5 queue?

 trying to imagine how it could make a team of five when some groups are two, and some would be four.

 

Yeah. If you remember early on there was a season or 2 that was TeamQ only. You could only queue in a party of 2, 3, or 5.

 2, 3, or 5 are the best option. If you have exactly 4 people, you may as well just grab 1 more and then queue with a full team.

And SoloQ Ranked should be SoloQ only.

Literally everyone who didn't piggyback off DuoQ to get top ranks every season would be happy.

For the 20ish people that did: 😂👉🚪

[SoftFootpaws.9134]

This won't change anything because the default matchmaking goes way beyond 100 points of difference now, due to adjustments over time from low player population. I imagine its even worse now with most bots gone and frequent festivals.

 

You already get regular games where everyone's solo q'd but still seem to have 500, even 1k skill difference. I killed an entire team the other day in ranked and I just don't know whats going on anymore. 😕

 

As I said in another thread, if anything unranked is more stable at this point. Though, alot of players would rush to defend 5man queue as being responsible for this, but I don't think that's the case. I think ranked (and PvP in general) population is just too low for MMR to do anything significant at this point.

 

You could just match up players completely randomly and still score a 50% average win/loss rate.

Edited December 31, 2021 by Hannelore.8153

[Luthan.5236]

Shouldn't the rating system actually take care of it? I mean if 1600 queues with 1000 and the matchmaker uses the 1600 and matches to 3 others with 1600 it would be (1600+1600+1600+1600+1000)/5 for the other team. (Afaik only the own rating is compared vs the average/mean of the enemy team.)

Meaning they'd lose more (if they all were 1600 ... might be lower if the matchmaker also tries to make a close mean/average score for both teams) but also would have it easier to win cause everyone would farm the 1000-ish guy.

But now if he really was fake rating (real skill at 1600) that team would be more likely to win and with his 1000 compared vs. the enemy team his score should rise quickly so he won't stay at 1000 forever.

Edited December 31, 2021 by Luthan.5236

[shion.2084]

20 hours ago, lightstalker.1498 said:

 

Nah. This is not true. It looks at each person individually, and builds a team average, regardless if they are in a duo or not. It doesn't even try to put one duo on each team, that is why you have some games where one team has two duos and the other has none.

 

That should be proof enough, but in case it isn't, here is the configuration.

 

<Arena name="Ranked Arena">
  <Queue>
    <RosterSize min="1" max="2"/>
    <Iteration interval="30s" rosters="100" limit="250ms"/>
    <Potentials min="20" max="500" falloff="0.375" start="1m" end="3m"/>
    <Rating start="5m" end="10m" max="1200" min="25"/>
    <Power curve="1" percent="1"/>
    <Rank min="20"/>
  </Queue>
  <Matcher type="Team">
    <Age seconds="2"/>
    <RosterSize max-diff="3" distance="-100" perfect-fit="0"/>
    <Rank distance="0"/>
    <Rating distance="-10"/>
    <Profession max="2" common="-100" unique="0" matching="100"/>
    <Dishonor distance="-100" stack="-50"/>
    <GuildTeam affinity="50"/>
    <Games max="500" distance="-0.25"/>
  </Matcher>
</Arena>

 

Notice how is is looking at the individual, regardless of their queueu size, it does this for each player so it can try to build a fair match of players close to the same caliber. If I play at the plat level and I queue up with a guildie who is at the lowest silver tier, I definetely can see a difference in the play.

I have played in games with 5 necrosis, where none switched class…. Is “profession” by specific type of necro?  Otherwise this algorithm isn’t what is usef

[shion.2084]

Also, I always thought I’m ranked the score of a duo was calculated as the higher of the pairing to stop gaming the system…. And this is what it should be

[shion.2084]

If you allowed the duo to average, then you could use fake dog accounts ( from win trading) to duo with the friend at the end of the season and have two players that are better than they should be boost the higher player.  So that would be idiotic 

[Genesis.5169]

Just make it so you dont lose medals so i can play with my silver friends and tank to g1 and not silver 1 with them cause they so bad.

Edited December 31, 2021 by Genesis.5169

[lightstalker.1498]

7 hours ago, shion.2084 said:

If you allowed the duo to average, then you could use fake dog accounts ( from win trading) to duo with the friend at the end of the season and have two players that are better than they should be boost the higher player.  So that would be idiotic 

 

It is not taking the top or the average. It looks at each player individually in the duo, regardless. Teams are average, and built by each player, even if they are in a duo. Players do play with dog accounts for a boost. There are well known dog alts...

[Bazsi.2734]

On 12/30/2021 at 10:17 AM, lightstalker.1498 said:

It's not friends playing together, it's a person who wants to climb with a buddy on a low rated alt to get easier matches. 1600-1700 or higher rated player queues with a false silver, who is currently rated in the 1000s or 1100s. You know what kind of match that gets them. Thing is that same low rated account loses alot of matches, but not when they duo with the high rated buddy.

 

Suggestion is to lock up duos to 100 rating points or less disparity. Matches will be more fair, and you will see less wacky stuff in games. I don't want duos to go away, but players need to queue with others of the same caliber and not try to manipulate for easier matches, you know, I know, we all know this is cheating even if  (nerd voice) "technically" it's allowed. Time to make it "not allowed".

 

Besides if only so many people duo, then you will see less bullcrap duo's in your games, and have the odd silver rated holo or herald named FHkjshdy, jabbarooing 3 of your team mates while you are wondering what the hell is going on. Limit the gap to 100 rating points or less.

 

Easy fix.

Nah, pass. Imagine you have a friend with whom you sometimes duo... you skip a day, and now you cannot duoQ anymore because he won/lost 5 matches yesterday. It would create this weird world where some people are allowed to play with friends, while others are not.
 

If your goal is improving match quality, just outright delete duoQ, solving most of the unfairness matchmaking has to deal with.

[Trepidation Lost.3469]

My first game of the year was vs a duo who both had that trophy floating above their head and another on their team had got of pvp. 
so yes, I agree, matchmaking is kitten. 
 

[shion.2084]

1 hour ago, lightstalker.1498 said:

 

It is not taking the top or the average. It looks at each player individually in the duo, regardless. Teams are average, and built by each player, even if they are in a duo. Players do play with dog accounts for a boost. There are well known dog alts...

Source?  
 

as pointed out the pseudo code was ambiguous.  I just can’t see a fellow developer setting up the math that way, knowing how it would get abused.

[lightstalker.1498]

3 hours ago, Bazsi.2734 said:

Nah, pass. Imagine you have a friend with whom you sometimes duo... you skip a day, and now you cannot duoQ anymore because he won/lost 5 matches yesterday. It would create this weird world where some people are allowed to play with friends, while others are not.
 

If your goal is improving match quality, just outright delete duoQ, solving most of the unfairness matchmaking has to deal with.

Sounds good. Delete duos. Easiest fix. And remove class swapping too.

[Genesis.5169]

7 hours ago, lightstalker.1498 said:

 

It is not taking the top or the average. It looks at each player individually in the duo, regardless. Teams are average, and built by each player, even if they are in a duo. Players do play with dog accounts for a boost. There are well known dog alts...

You remove duo you remove spvp, many will leave out because of that fact, I am one of them i and i know atleast 10 plat+ players who will leave over this. Cutting more content is not the way.