Guild Wars 2

According to Anet, this is how the matchmaking system works right now:

https://forum-en.gw2archive.eu/forum/game/pvp/High-MMR-is-punished-for-solo-que/first#post5000108

The current system is more likely to produce a result like this…
Red: 10, 5, 4, 3, 2
Blue: 6, 4, 3, 2, 1

Which could be argued is less fair, but really because MMR isn’t the only factor this is an oversimplification.

My suggestion is that for pug vs pug, it should be looking at multiple matches so if we have players of random rankings between 1 and 10, the matches should look like this:

Red: 1,1,2,3,2
Blue: 1,3,1,3,2

Red: 4,5,7,5,6
Blue: 7,7,4,4,6

Red: 10,9,9,8,8
Blue: 8,8,10,10,9

The game prioritize matchmaking in the following way:

1. Put people of similar levels into the same match. A 1 should never play against a 10.
2. After doing #1, try to make the match as even as possible.

If it’s a pug vs a premade, the average MMR of the premade team should be multiplied by something like 1.15x or 1.25x in order to take into account the obvious advantages a premade has versus a pug. Anet can look at the point distributions and adjust this multiplier as needed in order to not have premades dominate pugs but also vice versa.

The reason I say point distributions is because W/L ratio is bull. Someone can have a 50/50 W/L ratio but if every game is 500 to 100 either for or against, this is not good matchmaking and not fun pvp. It’s much more important to have games be 450 to 500 and maybe the W/L strays from a 50% average because a competitive match that is lost is much less disheartening than a blow out loss.

I know people will say “this is exactly what will happen in leagues”. Given Anet’s history with matchmaking and their refusal to address the inherent imbalance of soloq vs premade q, I am very skeptical.

Finally, someone will inevitably say “they can’t do what you’re suggesting because populations are too low”. This is untrue according to Anet: ( https://forum-en.gw2archive.eu/forum/game/pvp/Leagues-1/first#post5511329 )

hi Torso, your suggestions seem reasonable. What’s funny is that before Evan’s announcement, we had people defending matchmaking saying it was working as intended. Oh dear my word. Im not sure how they will resolve these bugs but I’m partially relieved that they’re aware of it.

TorsoReaper.8530:

If it’s a pug vs a premade, the average MMR of the premade team should be multiplied by something like 1.15x or 1.25x in order to take into account the obvious advantages a premade has versus a pug. Anet can look at the point distributions and adjust this multiplier as needed in order to not have premades dominate pugs but also vice versa.

The reason I say point distributions is because W/L ratio is bull. Someone can have a 50/50 W/L ratio but if every game is 500 to 100 either for or against, this is not good matchmaking and not fun pvp. It’s much more important to have games be 450 to 500 and maybe the W/L strays from a 50% average because a competitive match that is lost is much less disheartening than a blow out loss.

They already give an additional 2% (forgot the actual numbers) for premade vs solo queuers. The “odds of winning” increases for the premades. That meams that the soloQ team wont receive that bad of a loss in points because of the odds being against them.

Anet said that it’s hard to govern exactly how much of a “buffer” to give solo players vs premade matchup because every match, every team, is different.
Personal experience: I’ve beaten several premades while on a solo team. I just wish I could see my “odds of winning” in a match.

Furthermore, Anet said that the average points a losing team can reach are around 300. If a player consistently reach a certain point threshold, it’s hard to determine exactly how “good” this player is.
(In other words, if the majority of solo players have an average of 300 point distribution, then it’s difficult to differentiate mmr amongst them)

It’s why there are many factors govering MMR matchmaking, even ones we may not know. Anet just gave us a simplified version of how it works.

My point isn’t how you calculate MMR, it’s how you match people based on their MMR.

Right now it matches people who don’t even know basic capture point rotation with people who have played over 1,000 matches. This doesn’t make any sense. A person who is very good shouldn’t get weighed down by a team of new players and a new player shouldn’t have to fight against people who are exponentially better.

If Anet is doing 2% they need to raise it. You clearly disagree that premades are a problem. Maybe you only play premade and that’s why you defend the status quo so vociferously. But at the end of the day, matchmaking has been markedly worse since they took away the solo queue option.

Considering onyl your first post, thats what the alogrithm does (despite that hightier-lowtier bug). The example you quote was an answer to a post before where it was a boundary condition to get those exact playsers into one match. The only question was how to distribute them.
Of course the algorithm tries to mix players with similar mmr. Thats why they fix this bug that caused to break this rule.

Tux kitten inguin.3049:

Considering onyl your first post, thats what the alogrithm does (despite that hightier-lowtier bug). The example you quote was an answer to a post before where it was a boundary condition to get those exact playsers into one match. The only question was how to distribute them.
Of course the algorithm tries to mix players with similar mmr. Thats why they fix this bug that caused to break this rule.

I guess we will see on tuesday but that isn’t how i read the patch notes. I read it as sometimes within the teams they formed, the best player gets put on an extra bad team. Not as in “we put all the new people together in one match and veterans in another”.

If you can tell me you’ve never played with people who don’t even know how to cap points, then maybe I will believe you but we all know there are some serious issues with people who literally know nothing about PVP being put into matches they have no business being put into and it hurts everyone involved.

The idea to give to premades a little multiplier on their average mmr is not bad at all.

I like it

Read the final post in this thread.
Concerning proper distribution of MMRs amongst three node play.
These are the two biggest core problems plaguing match making.

https://forum-en.gw2archive.eu/forum/game/pvp/High-MMR-is-punished-for-solo-que/page/3#post5540416

It scrolled off the first page of the forum too quickly today. It should have stuck around long enough for people to read.

Trevor Boyer.6524:

Read the final post in this thread.
Concerning proper distribution of MMRs amongst three node play.
These are the two biggest core problems plaguing match making.

https://forum-en.gw2archive.eu/forum/game/pvp/High-MMR-is-punished-for-solo-que/page/3#post5540416

It scrolled off the first page of the forum too quickly today. It should have stuck around long enough for people to read.

I completely agree with almost everything you wrote. I still think 3’s should not be with 10’s (in the theoretical sense that we have exact mmr), but if the pvp pop on a certain day was only 10 total people, I think your suggestion is the best possible team balance.

I would much rather have a longer queue time than getting put with a terrible team and having a guaranteed loss. Long queue time of 5 minutes to get into a good game = 5 wasted minutes. Short queue time of 2 minutes to get into a 15 minute 100 to 500 blowout = 17 wasted minutes.