Jump to content
  • Sign Up

How does ranked matchmaking work for teams?


ixora.3569

Recommended Posts

Do they give you teammates that are same skill level? because I was matched with people who said they were new and didnt know how to play. So I wasnt sure whats going on here while my teammates were still learning what to do the other team just raped us. I dont mind playing with new players I just dont think it was fair to match us with people who know what they are doing. Maybe I was being trolled iono it just wasnt fun and a major waste of time when its 501 - 25.

Link to comment
Share on other sites

up to gold 2 you get even matches in terms of matchmaking

once you hit gold 3 you can be placed with top 10 players, then it's just a matter of lucky if you playing with or against them.

If you playing with them you prolly gonna win unless you afk, your chances are high even afk.

If you playing against them you can play the best game of your life and your chances are like 10%

Link to comment
Share on other sites

Depending on the daytime you play, you can have high platinum players in the same match as low silver players.

The match making is a joke and might as well be reducing to simply collecting any 10 players and randomly put them in teams.Other than faster queue times, the result would likely not change from what we have now.

Link to comment
Share on other sites

New players get placed on 1200 mmr if im not mistaken, so if you sit around 1100-1300 it shouldnt be to uncommon to play with new players considering the volatility of new accounts and the pretty low overall playerbase. Even 1400 should possibly be able to encounter new players rather easily if they had a few lucky wins.

Link to comment
Share on other sites

@"ixora.3569" said:Do they give you teammates that are same skill level?

No that's not how matchmaking works. At least usually, I actually have no idea about gw2 specifics but a matchmaking system (mms) usually tries to balance the average of teams with the least deviation as possible.The longer you have to wait the more deviation the mms allows.So assume 2 new buddies with an mmr of 1200 (default).If the 2 players are looking together for a 2v2 match, then ideally the mms would match them against 2 other 1200mmr players. The average mmr of both teams being 1200 with 0 deviation - perfect scenario.But if the amount of players searching for a 2v2 match is abysmally low and you have to wait for an eternity (so that the mms increases the allowed deviation by a lot) then it would still be perfectly fine (tho not desirable - even for the mms) to match them against a 1600mmr and a 800mmr player (the average of both teams being (1600 + 800) / 2 = 1200 again). In theory the advantage of the 1600mmr player should be canceled out by the disadvantage of the 800mmr player. Depending on a number of things this can actually be pretty acurate or not even remotely.On a side note: Your skill rating gain and loss after a match depends on your personal mmr vs the team/opponent's average if I remember correctly, so the 1600mmr player in the scenario above will lose A LOT of rating if they lose and the 800mmr player will lose barely anything. The opposite is true if they win, so the 1600mmr player will gain little to no skill rating (some Legendary players even got +0 in some matches) but the 800mmr player will gain a lot. So if you get like +11 to +13 rating at the end of a match then you were pretty well balanced with/against the team (purely by rating, there's a lot more to gw2 matchups imho tho so it can still feel very one-sided, for example the team compositions).If you gain like +8 or something this means you were meant to win that match due to the enemy team's average mmr being a lot lower than your personal one so it "should be easy anyway". On the other hand if you gain like +17 it means the enemies were actually supposed to win based on mmr (aka their average is higher than your rating). At least if the volatility has worn off.

Why 1200mmr as default you may ask? Because that's usually ~about the average mmr across all players, most people are somewhere in gold1-3 if I'm not mistaken. Due to the volatility you get "corrected" VERY quick.. Like.. lose your very first match against a team with lower mmr and you'll lose ~100 rating (or gain 100 if the numbers are right and you win). But volatility is a whole other topic.

So if you are somewhere in gold you are very likely to get matched with(or against) new players from time to time. And even if you are high silver/low plat you will have some completely new players when queue times are long. That's also the reason why I always cancel and restart the queue if it takes longer than ~4-5 minutes.I have no idea whether or not this helps tho or if the gw2 matchmaker works differently regarding the deviation and 4 minutes is also a completely arbitrary number.I'd also prefer to start out at ~1100 or maybe even less mmr. I don't know for sure if this is also tied to the re-placement matches but if you have mmr from a previous season then you will... I don't know if I remember correctly.. Start with (1200 + [Last Season MMR]) / 2 mmr and then carry on with the 10 placement matches (which only hide your gain/loss btw, they still count like every other match with A LOT of volatility).

Link to comment
Share on other sites

@"DoomNexus.5324" said:

Just to abridge what you said, how the matchmaker works for a single player, is the same as how it works for duo players, by just making the duo be treated as a single player in the eyes of the matchmaker (By averaging their MMR)

They have the code here on the wiki : https://wiki.guildwars2.com/wiki/PvP_Matchmaking_Algorithm

The code is written by the dev who designed it. Lots of interesting pieces of information for people who can read and understand the code. The MMR is only half of the process in terms of matchmaking, the other half, being which team the matchmaker decides to put you on. Using both it can evaluate what team a player should be on. If one were to ask me, mathematically it doesn't make sense to even have a system like this in place for a multiplayer game mode. The system it's based on (Glicko) was designed for 1 vs 1 player chess games, where it makes sense because no other person can statistically change your rating. However in a multiplayer setting, your rating is essentially as good as the average rating of your team, and where the win condition is simply to win or lose, means MMR which is supposed to be "defined" by player skill is not defined by your actual skill level, but rather your teams average chance to win or lose a game. That to me is the huge problem with the system as a whole, and also why we see such DRASTIC player skill level differences from player to play in g1-g3, where statistically is supposed to be the top of the populations bell curve for skill level, and yet brand new players and veterans are placed there.

I could go on in depth about why the above is a terrible system with some hardcore math's but I've lost my enthusiasm for trying to show it to others. It's just a downhill battle and nothing ever seems to change so why bother u know what i mean.

Link to comment
Share on other sites

@JusticeRetroHunter.7684 said:

@"DoomNexus.5324" said:

Very interesting, I didn't know this was public. Thanks for sharing.

The system it's based on (Glicko) was designed for 1 vs 1 player chess games, where it makes sense because no other person can statistically change your rating. However in a multiplayer setting, your rating is essentially as good as the average rating of your team, and where the win condition is simply to win or lose, means MMR which is supposed to be "defined" by player skill is not defined by your actual skill level, but rather your teams average chance to win or lose a game. That to me is the huge problem with the system as a whole, and also why we see such DRASTIC player skill level differences from player to play in g1-g3, where statistically is supposed to be the top of the populations bell curve for skill level, and yet brand new players and veterans are placed there.

I partially agree. I don't think Glicko is not suitable for multiplayer games in general. CSGO for example uses glicko-2 as its core rating system and I think it works really well there. I just think that gw2's way of determining rating gain/loss is abysmal. For example if - for the matchmaking system - my high rating is in theory offset by lower mates then I shouldn't get punished more than them if we lose imho aka we should all lose or gain the same amount of points. For me that just doesn't make sense from a fairness point of view. I understand that this is to prevent people from breaking through the roof or go to 0 rating because growth and decline wouldn't be bound to the opponents but this makes it really hard for certain classes imho or at least harder than for others. Some classes just can't carry as much as others can so just because I have a very high rating compared to my enemies doesn't mean that I can win the game single-handedly. Of course it also depends on my team's and my enemy's team comp etc.

Stamping a player with a single number is a bit problematic in general imo, since people vary from day to day and their matchups are also not equal. Sometimes I have matches where I get absolutely obliterated on T side for example but then on CT I get like 2 kills per round average and clutch rounds in 1vX. Maybe it has to do with the team eco and I'm not that good when we have to do halfbuys or ecos every other round or so.In GW2 for example classes, elite specs and even down to certrain builds will matchup very differently and I think this should be considered for the matchmaker as well. But then again I understand that this will probably never happen since the MM would need to be reconfigured every time they shift the balance and other (un)favourable matchups emerge.

But I'd love to see GW2 take a step towards the csgo model (or literally any other competitive game really) to first of all hide the actual mmr and only display the skill group. Also Anet pls make the gain and rating a bit more elaborate than what your personal rating is compared to the average. There are a lot of metrics available and while none is really good at representing how big your impact in the game was and therefore how you performed, it sure will be more representative if several metrics are used than just say "ok your rating was 50 above average now you get only +9" or something.

Link to comment
Share on other sites

a year or so ago, it went based on the skill level of the lowest pvp leveled person in your team.So you'd get really weird games, like the one where we were up against the level 1 person & gold-plats playing with their first time pvp playing friend, while we were playing with our first time pvp playing friend XD

i don't mind the goofiness of it though.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...