Guild Wars 2

There were several errors with the way the MMR algorithm is being applied this pvp season:
1. All new players were brought in at average MMR, making it impossible for the algorithm to distinguish players who were actually average from ones that are below average, but new.
2. MMR is used to determine your teammates, but not your opponents. This actually reverses the purpose of MMR. Rather than finding you opponents of the same skill level it becomes a self-fulfilling prophecy. You have low MMR, so you get stuck on bad teams, so you lose and your MMR drops further.

These two decisions together have resulted in a significant group of the average players getting stuck in vicious circle.

At the beginning of the season, they got stuck with a bunch of new players and lost (because there is no way to demonstrate your ability apart from your team). After that first game, all the new players and the average players stuck with them dropped a chunk of MMR, while all the average players stuck with experienced players gained a chunk.

In the second game, the average players that dropped are even more likely to get stuck with new players, so they lose again and drop another chunk. After a couple of games, they’re now doomed to always be with a bunch of new players, unable to ever demonstrate competence because they have no way of ever qualifying to play with better teammates.

Yea spot on. Thats exactly how it works right now.

The thing with this system is it will balance out in the end. At a certain time the lower MMR players will face eachother because they are the only people within the pip/division search range. HOWEVER this might take a decent chunk of time to balance out. If it takes too long to calibrate everyone, then this system is totally not worth it because you will spend way too much time playing garbage quality games.

I still prefer a balanced team vs balanced team setup over this current setup.

1. what if new players start off with zero or way below average MMR instead of average MMR?

2. this is fine really, only 3 outcomes, you versus a team with higher, lower or similar MMR.

btw average players can get together, form a casual team, then go stomp newer players and rise back up from the ashes of defeat to be victorious!

That’s the thing, it won’t actually balance out ever. Once you’re segregated into that “low MMR pool”, you’ll always be in the “low MMR pool”. You may eventually migrate out of Amber into Emerald, but only at the rate the rest of those bad players are migrating out, so you’ll still be stuck playing with them in Emerald.

The fundamental problem is they’re using MMR backwards. It’s purpose is to rate two teams to see if they should compete and then update the ratings after the match based on the result of that competition.

It’s not meant to be used to form those teams to begin with. Since you don’t compete against the rest of your team, there’s no “competition” to meaningfully update the rankings at the match.

Oh Snapalope.1378:

The thing with this system is it will balance out in the end.

Even tho, this does not justify a week of bad matches.

Please take a look how the league system in sc2, lol or any other game are working. They still manage to fulfill all the pros of the league system without suffering through a period of bad games. (E.g. Just take a look at my other posts, I have multiple times explained how the league of legends system is working.)

Deimos Tel Arin.7391:

btw average players can get together, form a casual team, then go stomp newer players and rise back up from the ashes of defeat to be victorious!

You’re right and this may be the only way out of the vicious circle. But should you have to be on a premade to compete in ranked play?

Oh Snapalope.1378:

The thing with this system is it will balance out in the end. At a certain time the lower MMR players will face eachother because they are the only people within the pip/division search range. HOWEVER this might take a decent chunk of time to balance out. If it takes too long to calibrate everyone, then this system is totally not worth it because you will spend way too much time playing garbage quality games.

I still prefer a balanced team vs balanced team setup over this current setup.

The problem is it still won’t be accurate. There will be people with lower MMR than their actual MMR and there will be people with higher MMR than their actual MMR, further increasing the inaccuracy this system brings.

Anyhow, this is the first game I’ve played where you start playing the game and get an average MMR. In other games, new players start at the bottom/calibrate. It’s ridiculous to have this and then not consider MMR of both teams (never seen this either). I’m not sure what gw2 is trying to create, but with every pvp season it gets worse and worse.

Every time I play against the same persons in solo queue, every time! That’s normal?! I’m winning, sure, but it’s frustrating! Really! If one day I’m logged in I play against and with those players all day… now try to advance! I will for sure, but it will be the same in ruby to, than u are really annoying me!

Oh Snapalope.1378:

The thing with this system is it will balance out in the end. At a certain time the lower MMR players will face eachother because they are the only people within the pip/division search range. HOWEVER this might take a decent chunk of time to balance out. If it takes too long to calibrate everyone, then this system is totally not worth it because you will spend way too much time playing garbage quality games.

I still prefer a balanced team vs balanced team setup over this current setup.

That’s not accurate. There’s going to be wayyyy too much variance among players with the same skill level to make that claim. The system isn’t necessarily going to move good people up and it’s there’s not even good probability of it moving the best players up to the high tier. It’s going to place people. That’s really all you can say.

I think the thing that enrages me the most about this season is that it is like a slap in the face to statistics and math in general. It is genuinely insulting the lack of background they feel they need.

Laraley.7695:

I’m not sure what gw2 is trying to create, but with every pvp season it gets worse and worse.

To be honest, it’s what happens when people without a statistics background take a cookie-cutter statistics algorithm they don’t fully understand and try to apply it in a way that seems superficially similar to its intended use, but which actually violates a number of its assumptions.

MMR is used to determine your teammates, but not your opponents.

Okay when I first started the season I didn’t really read up on the changes to matchmaking but now that I have; this seems like the single worst decision you could possibly make for a match making system. Why doesn’t the system try to find two matching average MMRs between the two teams. It makes absolutely no sense to not try to match the two team’s MMR’s. This is leading to an enormous amount of lopsided games.

Oh Snapalope.1378:

The thing with this system is it will balance out in the end. At a certain time the lower MMR players will face eachother because they are the only people within the pip/division search range. HOWEVER this might take a decent chunk of time to balance out. If it takes too long to calibrate everyone, then this system is totally not worth it because you will spend way too much time playing garbage quality games.

I still prefer a balanced team vs balanced team setup over this current setup.

Honestly i really didn’t like struggling through amber and emerald because they kept trying to put me in fair matches. Even if it means i get hard capped eventually because of my lack of skill at like Ruby or something, I’d still rather take this system where i can streak right past the lower divisions

Stormy Dragon.9210:

Laraley.7695:

I’m not sure what gw2 is trying to create, but with every pvp season it gets worse and worse.

To be honest, it’s what happens when people without a statistics background take a cookie-cutter statistics algorithm they don’t fully understand and try to apply it in a way that seems superficially similar to its intended use, but which actually violates a number of its assumptions.

YESSSS the devs linked a normal distribution in a post and I wanted to scream!!!!!!!! Part of me thought surely the comment must be a joke.. but there was just no way.

This is actually very interesting to see something that works fantastic in theory but, sucks horse’s rear in reality.

Essentially, season 2 is all about attrition. For matches to be semi-decent 2 types of attrition need to happen, not one.

One – People with higher skill sets (let’s say 60% or more winrate will have to move far away from the PiP range of players with 60% lossrate) and eventually stop playing for mid-range to climb up.

Two – (No one talks about it) Several people with 60% lossrate need to stop playing as they can’t move out of the bracket of losers to make the games of losers at least decent (not rushing beast with 4 people at the start or fighting to cap ferocity etc.). This is based on assumption that lower skilled players will eventually say “kitten u anet I am done with your kitten” and throw the towel. Then the bracket of losers (Sapphire and below) will have semi-decent matches.

Attrition at higher level will have to happen as many people will not go past diamond t1 and matches will take forever unless ANET chooses to pit the legendary “winners” against the “losers” to keep queue time in check for the winners. The rate at which people hitting ruby (3 days into the league), I’d say attrition at legendary will be extremely quick because for most of those winners there is nothing to gain this season after that point or one more pass thru the legendary pool.

Attrition in lower level will happen within a week or less. Based on the sudden lack of enthusiasm in league from previous legends that I know, I am pretty sure some of them realized they would not go past sapphire this time and stopped playing as soon as they collected the tickets for certificates. For other bottom feeders… well 2 weeks should be enough.

Now, if all the people actually stick around during the entire season then it will reek like Reek. I am sure ANET is betting against this to happen.