Guild Wars 2

Lots of people seem to be confused on how things are working. Depending on your situation, your expected win % can vary a lot.

MMR – hidden rating that moves around and tries to estimate your skill level
Pips – reward rating that measures a mix of your effort and your skill level

Matchmaking will find players who are within 15 pips of each other, then order them sequentially by MMR, putting the top 5 players on one team and the bottom 5 on the other. This ordering is done by MMR, not by pips, so a team with lower league players could be the higher MMR team; you really can’t tell if you’re on the high or low MMR team (so don’t worry about it).

Here are some scenarios:
- You are more skilled than players around your pip level – You win more than 50% and climb fast
- You have an equal skill level to players around your pip level – You win about 50%
- You have a lower skill level compared to your pip level – You win less than 50%

If you’re in the first group, ride the wave and be ready for when you hit the wall. If you’re in the second, keep playing and you’ll slowly gain pips thanks to streaks (and if you’re below ruby, tiers). If you’re in the third group, you need to focus on improving if you want your win rate to improve.

There’s also a secret scenario: Your MMR does not accurately represent your skill level at the moment. This can be due to being on different profession or build than you had been using, as well as natural gameplay quality variations. This scenario can and probably will apply to some number of players in any given game. If this is the case, then higher MMR team might actually be the weaker team and the lower MMR team the stronger.

At the end of the day, you should ride the win train until you hit the wall and then either focus on improving, keep playing if you’re having fun, or quit until next season. The system may slightly stack the teams, but unless someone is really far from where they should be pip wise, it won’t matter much.

wolfer.1587:

Matchmaking will find players who are within 15 pips of each other, then order them sequentially by MMR, putting the top 5 players on one team and the bottom 5 on the other. This ordering is done by MMR, not by pips, so a team with lower league players could be the higher MMR team; you really can’t tell if you’re on the high or low MMR team (so don’t worry about it).

It will try to find players who are within 15 pips and a certain MMR range - otherwise it would be pretty much the same as season 2.

What i wonder – does the system really searches for 10 players and then divides them into 2 teams or does it create teams out of 5 players and then match those teams against each other?

wolfer.1587:

If you’re in the third group, you need to focus on improving if you want your win rate to improve.

Or just wait, until the majority of better player has moved on. The longer the season is going, the easier climbing will become.

What you describe is the season 2 matchmaking. Season 3 apparently has changed this, although to which extent we do not know (but it’s safe to assume it’s not longer top5 mmr vs bottom 5 mmr):

Season 2 matchmaking was changed from Season 1 Ranked and Unranked matchmaking to help ensure league positions correlated more with skill instead of simply being based on the hours played. These changes saw a multitude of improvements for Season 2, but unfortunately they also spread out win rates further than we would have liked, as matches became a bit too uneven. We’ve created a new matchmaker for Season 3 that will strike a balance between the two previous seasons, accomplishing the same goals of accuracy from Season 2 but doing so with the evenness of matches that players experienced in Season 1.

Well, I am getting paired with people who don’t fully cap home/mid just so they can follow some guys across the whole map. With poeple who zerg far WHILE home/mid is getting capped by a thief. I don’t believe someone like that is within my skill level, so why does the matchmaker keep on pairing me with that match after match?

wolfer.1587:

There’s also a secret scenario: Your MMR does not accurately represent your skill level at the moment. This can be due to being on different profession or build than you had been using, as well as natural gameplay quality variations. This scenario can and probably will apply to some number of players in any given game. If this is the case, then higher MMR team might actually be the weaker team and the lower MMR team the stronger.

I don’t know about secret scenario, it may be permanent. I have to duoQ with diamonds to get anything done. In solo, I get an AFK or disco about 25% of the time. It has not yet happened to the enemy team for me. My win rate at the bottom of Ruby as solo is about 20% right now, while in duo/party with diamond is roughly 70%. I may not be losing pips for all these discos but I suspect it’s having a negative effect on my MMR.

I’ve been keeping logs on paper/excel and have nearly 500 gigs of video to help me study my own personal statistics. I want to see what personally contributes towards my progress, both in terms of skill and in terms of matching manipulation. Back in season 2, I had an even more drastic problem. Of all the matches I went SoloQ, I won only 11%, in duo 55%, and 4-5 man 65%. Once I hit legendary, my soloQ winrate climbed drastically, probably because there just wasn’t enough garbage present to hold me back. In season 1, gw2pvp.de ranked me in top 2000 for ranked play, which plummeted drastically for season 2. The line graph goes almost straight down. My observation tells me this pip system is hugely flawed somehow.

Maybe this MMR and Pip system is ideally good enough, but how MMR itself is calculated may be the actual issue. As I understand, it was intended to evaluate players based on an equal 50% skill split between the two teams. If you win, it’s because you may have been better than the estimated 50/50 split, thus adjusts it upwards. However if the teams are not split as such, losing still drops MMR. They admit it doesn’t affect it as much, but it still effects it enough to make an unusual win against prediction too rare to have positive results. Especially considering there are hard coded walls (the pip/division loss prevention) that completely inhibit recalibration. This should be resulting in runaway MMR issues (elo hell?).

Frankly, I just stick with joining parties. It hasn’t been much of a problem as I think the matchmaking has no choice but to drop MMR considerations and rely only on pip ranges when queuing as a 5 man party. My team mates feel I contribute much more than the random pugs they get and I have been showing much better performance considering my teammates are always where I expect them to be and doing what I expect them to be doing.

wolfer.1587:

There’s also a secret scenario: Your MMR does not accurately represent your skill level at the moment.

That’s not a secret scenario. Given the mechanics of the s3 system, that’s practically a given. Which, by the way, makes the first part of your thoughts pretty much irrelevant when applied to current season.

SneakyTouchy.6043:

Maybe this MMR and Pip system is ideally good enough, but how MMR itself is calculated may be the actual issue. As I understand, it was intended to evaluate players based on an equal 50% skill split between the two teams. If you win, it’s because you may have been better than the estimated 50/50 split, thus adjusts it upwards. However if the teams are not split as such, losing still drops MMR. They admit it doesn’t affect it as much, but it still effects it enough to make an unusual win against prediction too rare to have positive results. Especially considering there are hard coded walls (the pip/division loss prevention) that completely inhibit recalibration. This should be resulting in runaway MMR issues (elo hell?).

Exactly. They are using 2 separate systems, that work perfectly fine when used alone, but that actively hinder each other when put together. To make things worse they add up some even more damaging mechanics like the system intentionally favoring one side.

The whole system is put together in such a way that each of the subcomponents depends on the other subcomponent to work fine, but at the same time it prevents that subcomponent from working right, by distorting the data it works with. The end result is a real mess.

Astralporing.1957:

wolfer.1587:

There’s also a secret scenario: Your MMR does not accurately represent your skill level at the moment.

That’s not a secret scenario. Given the mechanics of the s3 system, that’s practically a given. Which, by the way, makes the first part of your thoughts pretty much irrelevant when applied to current season.

SneakyTouchy.6043:

Maybe this MMR and Pip system is ideally good enough, but how MMR itself is calculated may be the actual issue. As I understand, it was intended to evaluate players based on an equal 50% skill split between the two teams. If you win, it’s because you may have been better than the estimated 50/50 split, thus adjusts it upwards. However if the teams are not split as such, losing still drops MMR. They admit it doesn’t affect it as much, but it still effects it enough to make an unusual win against prediction too rare to have positive results. Especially considering there are hard coded walls (the pip/division loss prevention) that completely inhibit recalibration. This should be resulting in runaway MMR issues (elo hell?).

Exactly. They are using 2 separate systems, that work perfectly fine when used alone, but that actively hinder each other when put together. To make things worse they add up some even more damaging mechanics like the system intentionally favoring one side.

The whole system is put together in such a way that each of the subcomponents depends on the other subcomponent to work fine, but at the same time it prevents that subcomponent from working right, by distorting the data it works with. The end result is a real mess.

Ha! Seems about right to me!

In chess tournaments ELO is used. While our MMR isn’t calculated the same way it has something similar. In chess if your ELO is 1500 and you play against player who has 2000+ ELO and you win your ELO rating gets a big boost because you are defendant in that match. If you lose, your ELO drops less than higher ELO players if he loses.

But as you advance in a ladder and get winstreak going you are going to face bigger and bigger MMR players. It doesnt mean those matches are guaranteed losses, it just means if enemy plays classes that they are familiar with and have equal teamcomp, they have bigger chance to win. Most of those 3rd or 4th games in winstreaks are these kind of matches (atleast for me) no matter do I soloQ or duoQ. I see from enemy movement, splits, skill uses and map coordination that they are equal/better skilled players. But the win isn’t guaranteed to those players. Because humans make mistakes. And those mistakes can costs them whole match if enemy recognizes them and uses them to their own advantage.

But what gets you to losing streak after losing win streak is your mentality. You just won 5 games and now lost 2? Take a break, breathe, calm down and try again. My guildmate got into diamond with this strategy in 2 days after he was stuck in T3 Ruby for 5 days. You have still a month to play this season so no need to hurry.

Mutaatti.2789:

In chess tournaments ELO is used. While our MMR isn’t calculated the same way it has something similar. In chess if your ELO is 1500 and you play against player who has 2000+ ELO and you win your ELO rating gets a big boost because you are defendant in that match. If you lose, your ELO drops less than higher ELO players if he loses.

But as you advance in a ladder and get winstreak going you are going to face bigger and bigger MMR players. It doesnt mean those matches are guaranteed losses, it just means if enemy plays classes that they are familiar with and have equal teamcomp, they have bigger chance to win. Most of those 3rd or 4th games in winstreaks are these kind of matches (atleast for me) no matter do I soloQ or duoQ. I see from enemy movement, splits, skill uses and map coordination that they are equal/better skilled players. But the win isn’t guaranteed to those players. Because humans make mistakes. And those mistakes can costs them whole match if enemy recognizes them and uses them to their own advantage.

But what gets you to losing streak after losing win streak is your mentality. You just won 5 games and now lost 2? Take a break, breathe, calm down and try again. My guildmate got into diamond with this strategy in 2 days after he was stuck in T3 Ruby for 5 days. You have still a month to play this season so no need to hurry.

This is why I record my matches, study them, and sometimes have others review them. Poor mentality causes us to spew boring old platitudes without considering evidence and recorded data – Blaming fault on what’s not at fault. It also demands we stroke our kittens any chance we get because something within us is broken and we have to fake success to lift ourselves back up. Poor mentality does not cause team-mates to lose connection or or accidentally throw their keyboard out the window every time they get near a point.

wolfer.1587:

Lots of people seem to be confused on how things are working. Depending on your situation, your expected win % can vary a lot.

MMR – hidden rating that moves around and tries to estimate your skill level
Pips – reward rating that measures a mix of your effort and your skill level

Matchmaking will find players who are within 15 pips of each other, then order them sequentially by MMR, putting the top 5 players on one team and the bottom 5 on the other. This ordering is done by MMR, not by pips, so a team with lower league players could be the higher MMR team; you really can’t tell if you’re on the high or low MMR team (so don’t worry about it).

Here are some scenarios:
- You are more skilled than players around your pip level – You win more than 50% and climb fast
- You have an equal skill level to players around your pip level – You win about 50%
- You have a lower skill level compared to your pip level – You win less than 50%

If you’re in the first group, ride the wave and be ready for when you hit the wall. If you’re in the second, keep playing and you’ll slowly gain pips thanks to streaks (and if you’re below ruby, tiers). If you’re in the third group, you need to focus on improving if you want your win rate to improve.

There’s also a secret scenario: Your MMR does not accurately represent your skill level at the moment. This can be due to being on different profession or build than you had been using, as well as natural gameplay quality variations. This scenario can and probably will apply to some number of players in any given game. If this is the case, then higher MMR team might actually be the weaker team and the lower MMR team the stronger.

At the end of the day, you should ride the win train until you hit the wall and then either focus on improving, keep playing if you’re having fun, or quit until next season. The system may slightly stack the teams, but unless someone is really far from where they should be pip wise, it won’t matter much.

this is a great post, but what if youre riding the wave and hit the wall, and that wall isnt you i find myself in the first scenario, i find myself playing my role well on my team not dying getting kills getting decaps (i play thief) not fighting off point u know all the p’s and q’s or whatever the term is. my problem is ever since i got halfway thru ruby i cant find a team where players dont over rotate or just plain get wiped and it costs me my hard earned pips to the point where im back in t1 and cant get out. Yes i want to get better at this game thats always part of why i play but its annoying when i cant because im stuck with players that dont know what theyre doing.