Guild Wars 2

How is the chance of gaining/losing pips calculated?

Are classes weighted somehow? Is legendary x1 the same of legendary x3? Is time played a factor, or is it only points scored? Is a 500-100 lost match that took 15m the same of a 500-100 loss that took 5m?

Which does need some tweaking to be honest…. That game gave me one PIP only =P

Skullkger.2804:

Which does need some tweaking to be honest…. That game gave me one PIP only =P

That’s cos you had duoq and they were all solo

That’s it.

btw, the meaning of censure names and chat? :d

I wish they’d scrap the algorith altogether and let good ppl win and bad ppl lose.

Evan Lesh.3295:

The algorithm is explained here:
http://wiki.guildwars2.com/wiki/PvP_Matchmaking_Algorithm#Match_Prediction

Translate please

Shirlias.8104:

That’s it.

btw, the meaning of censure names and chat? :d

You technically aren’t supposed to have names of other players in your screenshots/posts without their permission.

I won 4 games in row for 4 pips, then lost 2 games to lose 4 pips.

Gotta love that math.

Belial.9350:

Skullkger.2804:

Which does need some tweaking to be honest…. That game gave me one PIP only =P

That’s cos you had duoq and they were all solo

It was pre patch that changed the way duo quoing counts. So you think it’s fair that 4 diamonds and one ruby earn one pip for completely trashing 3 legendary 2 diamond? I don’t really get that. Was a game from last year.

Wouldn’t it be courtesy to let know people when the game starts if they are going to lose 2 pips or not, if they’re going to win 2 …

So no bad surprise => no unhappy people.

Evan Lesh.3295:

The average ladder position, rank, mmr, and max party size from each team is compared. The further away these are, up to the maximum specified in the configuration, the more skewed the odds.

Actually, you also need to take consideration of final team composition as they relate to current meta when 10 second counter starts to determine win chance. A Team with a composition closer to meta will always have a higher chance to win against a team that is not. I will give you an example, see screen attached.

Red is running a dual chrono, dh, rev, ele
Blue is running dbl ele, mes, rev, dh

This is a pretty close matchup and team composition at diamond level. With equal skill level, red will always outbunk/outcap blue w/ double portal and +1. It’s impossible to counter in this meta. Come February, probably both teams will suck with blue having a higher chance of winning because they will probably outdps red based on the information available. Now, this matchup is probably as close to meta comp as it can get but, still win chance is predictable as blue team folks knew within 30 seconds of the match who was going to win eventually. Meta was just stronger with Red. If you believe, taking class and team composition as a factor to predict win-chance is not kosher, the system will never even come close to be relevant and you will always face this issue unless of course in next season every match is a 10 necro matches like what we used to have back in 2013 terrormancing dhuumfire days.

Also, I think, PIP loss is a blunder and is completely not needed if the MM system works properly. With PIP loss, in many cases, ANET is punishing people for reasons beyond that person’s control (meta, achievement completion, disconnects, loss streak from poor matchmaking etc.).

Many people do not really care after certain point because the system gets tilted and that person loses so many matches at a stretch, all hope gets lost. Theoretically, one can’t lose a long series of matches at a stretch in this MM system. But, it does happen. I went from 7 off to diamond to 4 in Ruby and then came back to get to diamond with a series of wins. I played same class, same character throughout the season. So, there is something that’s unknown in the matchmaking process.

Scoring system needs to change and queue modes need to be segregated. It’s like trying to merge random arena with hall of heroes from GW1 perspective.

Evan Lesh.3295:

The average ladder position, rank, mmr, and max party size from each team is compared. The further away these are, up to the maximum specified in the configuration, the more skewed the odds. Each component counts for a different amount of the prediction. For instance, ladder position counts for 5/8th the prediction.

1) Is ladder position being calculated based on parties (e.g. legendary + amber partied calculates as 2 legendaries, or as 1 legendary and 1 amber)?

2) Can we please start seeing MMR of ourselves and our opponents so we can actually trust the results instead of the complete black box that is MMR?

Something just feels very very wrong with match prediction. More often than not, I’m losing multiple pips on very close matches AND in matches where there is 0% chance to win in the real world. Yet I almost never win multiple pips even when I’m on the team that dominates. And I have no way to tell why things are so messed up because normal players can see almost nothing of what is actually going into the predictions. And thus, I feel I am utterly wasting my time trying to advance in ruby – I win and gain one pip, then lose and lose 2 pips without there being any sane reason why. Heck, I’ve lost 2 pips on a TIED match that was perfectly even the entire match in terms of skill and ladder position and party sizes.

velmeister.4187:

Evan Lesh.3295:

The average ladder position, rank, mmr, and max party size from each team is compared. The further away these are, up to the maximum specified in the configuration, the more skewed the odds.

Actually, you also need to take consideration of final team composition as they relate to current meta when 10 second counter starts to determine win chance. A Team with a composition closer to meta will always have a higher chance to win against a team that is not. I will give you an example, see screen attached.

Red is running a dual chrono, dh, rev, ele
Blue is running dbl ele, mes, rev, dh

This is a pretty close matchup and team composition at diamond level. With equal skill level, red will always outbunk/outcap blue w/ double portal and +1. It’s impossible to counter in this meta. Come February, probably both teams will suck with blue having a higher chance of winning because they will probably outdps red based on the information available. Now, this matchup is probably as close to meta comp as it can get but, still win chance is predictable as blue team folks knew within 30 seconds of the match who was going to win eventually. Meta was just stronger with Red. If you believe, taking class and team composition as a factor to predict win-chance is not kosher, the system will never even come close to be relevant and you will always face this issue unless of course in next season every match is a 10 necro matches like what we used to have back in 2013 terrormancing dhuumfire days.

Also, I think, PIP loss is a blunder and is completely not needed if the MM system works properly. With PIP loss, in many cases, ANET is punishing people for reasons beyond that person’s control (meta, achievement completion, disconnects, loss streak from poor matchmaking etc.).

Many people do not really care after certain point because the system gets tilted and that person loses so many matches at a stretch, all hope gets lost. Theoretically, one can’t lose a long series of matches at a stretch in this MM system. But, it does happen. I went from 7 off to diamond to 4 in Ruby and then came back to get to diamond with a series of wins. I played same class, same character throughout the season. So, there is something that’s unknown in the matchmaking process.

Scoring system needs to change and queue modes need to be segregated. It’s like trying to merge random arena with hall of heroes from GW1 perspective.

Great way to explain it.

Taking from the code the mmr system only takes the average mmr rating of both teams (which have to be within 200 points) while wide making for quick que’s its really bad if you would have checked for personal rank with a diviation between the highest and lowest ranked player on a team that would have been way better, granted since anet hasnt reset the mmr rank at the start of the current system this extra code would have made for very long queues.

In short had this code been added and the mmr reset at the start of the season win loss would have stayed around 50% and matches would have been closer