Guild Wars 2

I’ve waited a week to post this, just to make sure I wasn’t simply having bad luck. Every single day, I lose 70-100% of the games I play during off-peak hours. During peak hours, my winrate seems to be about 50%.

At one point, my win-loss record was something like 70-70. Now, about 60 games later, I am 20 games under .500. The only thing that changed was a week spent playing during off-peak hours – something I had not been doing before.

I don’t have the time to walk everyone through Intro to Statistics, but trust me: I am not the problem. Unless you want to argue that my skill doubles or triples when the clock hits 8PM, my off-peak winrates are too many standard deviations from my peak rates. And both rates are too many standard deviations from my mean. Meaning it’s the system, not me.

edit – Obviously my numbers are approximate. I don’t keep a detailed record of my daily win distributions. But the .500 to 20 games under .500 is accurate.

I asked siri and google-now.

Their answer is because you are getting carried during prime time. When off-peak hours start there are not many people left to carry you =]

That or not enough people your level during offpeak hours and you get paired with higher end players.

kirito got that wrong.

During peak time you have balanced matches because you are paired with people of your ability against teams that are about as good as your are.

In off-peak times you are forced to try and carry very low MMR ranked people because there are not enough people of your skill available to play alongside you.

But it is essentially impossible to carry several poor players match after match, and your losses mount up.

kirito.4138:

I asked siri and google-now.

Their answer is because you are getting carried during prime time. When off-peak hours start there are not many people left to carry you =]

That or not enough people your level during offpeak hours and you get paired with higher end players.

You really want to stick with this theory? You don’t even have any reason to offer it, other than the fact it’s sarcastic and insulting to me.

Ignoring that it’s wrong, for quite a few reasons, it would still not even show that matchmaking is working. Your theory still says matchmaking is broken, you’re just saying it is broken in my favor.

Either I’m getting beaten down during off-peak hours or propped up during peak hours. There’s no version of this where matchmaking is working.

The match making has never ever worked in this game. Why pretend like it ever has and that it can be sorted. I often get in with rank 3 upwards to 60 odd. In no way is that “match making”.

Justin ODell.9517:

@zilcho.7624: Are you taking into consideration the rating reset that happened about a month ago?

1) Your MMR reset to a higher level than it was before. This should settle quickly because your deviation and volatility were also reset.

2) Playing during off-peak hours, especially while your deviation/volatility were high, could caused your MMR to settle in either higher or lower than it should be.

Of course, this is just speculation. I will keep an eye on your account and see if I learn more.

I think it might just be the player population. I get so many repeat players on my team, I’ve started recognizing players’ names. Having a player placed on my team three consecutive games is not uncommon.

Although, I thought the population was large enough – even during off-peak hours – for repetition to still be rare.

(Obviously, all of these repeated players are having the same issue I am. I would be the repeated player from their perspective.)

zilcho.7624:

Justin ODell.9517:

@zilcho.7624: Are you taking into consideration the rating reset that happened about a month ago?

1) Your MMR reset to a higher level than it was before. This should settle quickly because your deviation and volatility were also reset.

2) Playing during off-peak hours, especially while your deviation/volatility were high, could caused your MMR to settle in either higher or lower than it should be.

Of course, this is just speculation. I will keep an eye on your account and see if I learn more.

I think it might just be the player population. I get so many repeat players on my team, I’ve started recognizing players’ names. Having a player placed on my team three consecutive games is not uncommon.

Although, I thought the population was large enough – even during off-peak hours – for repetition to still be rare.

(Obviously, all of these repeated players are having the same issue I am. I would be the repeated player from their perspective.)

Repeated players are very common, and I honestly don’t see too much of an issue with them.

As for the problem you have: not trying to offend you but the only possible explanation I could have is that your MMR is very low, which causes you to be unable to find proper matches in off-peak hours due to not enough players being at or below your performance level.

If you winrate increases as the pool of participants decreases, that means you’re at the top layer of active players. Inversely, if it decreases at low activity hours, that would mean you’re in the lowest layer. Could this be the problem?

Gokil,

No, his MMR could be high, and he is forced to play with lower-skilled players on his team when that is all who are available in the queue.

Why are people so hostile here that they always choose to find some way to put down a poster, even when it is a stretch and they have to reach for the negative interpretation?

Justin ODell.9517:

@zilcho.7624: Are you taking into consideration the rating reset that happened about a month ago?

1) It could be that your MMR was reset higher than it was before. This should settle quickly because your deviation and volatility were also reset.

2) Playing during off-peak hours, especially while your deviation/volatility were high, could have caused your MMR to settle somewhere higher or lower than it should be.

Of course, this is just speculation. I will keep an eye on your account and see if I learn more.

There is currently an issue with the queue system. Inviting to roster seems bugged, maybe it is linked?

I have to agree with the op. Off-hour times are a nightmare for soloq. If you’re ranked fairly high, there are few people on in your MMR. This results in long que times, and then it’ll just throw you into some really low level games, that destroy your rating if you lose.

Gokil.2543:

Repeated players are very common, and I honestly don’t see too much of an issue with them.

It depends on the rating/matchmaking system, but generally having repeated players is very bad. You need randomness to smooth advantages and disadvantages out.

With random samples, you don’t have to worry about team comp, inaccurate ratings, or anything else. You are as likely to benefit from those factors as you are to suffer from them.

But if things aren’t random, suddenly that assumption doesn’t work. I don’t know how GW2 does it’s matchmaking, though. Repeated players may not break the assumption of randomness.

It also depends on what class you play, for example any tank class can carry a team. Many times the DPS will make the difference on two decent teams, but if you are playing during off peak hours prepare to plummet if you don’t play a carrying class (Bunker gerd, spirit ranger, any warrior [probably even 0/0/0/0/0 jk], possibly extended to engineer on far point assault or bomb nade).

Taym.8326:

… for example any tank class can carry a team.

I disagree. You can’t take and hold two points by yourself.

Silentshoes.1805:

Gokil,

No, his MMR could be high, and he is forced to play with lower-skilled players on his team when that is all who are available in the queue.

Why are people so hostile here that they always choose to find some way to put down a poster, even when it is a stretch and they have to reach for the negative interpretation?

As I said, I’m not being hostile. Logically, your explanation does not make sense. If he is forced to play with lower-skilled players, there will be 5 lower skilled players on the other team, and 4 on his team. Considering he is higher skill than them, he should be able to win more than 50% of the matches. I don’t understand why that is so hard to get. It’s the typical ‘ELO hell’ remark that people just don’t seem to want to let go of.

If on the other hand, OP is a below average player, and there are not enough people around his skill level, the lobby would fill with players of slightly higher skill, 4 of which would be on his team, and 5 of which on the other team. Result: OP’s team has a disadvantage and loses over 50 percent of his games.

This is just statistics. Judging by the fact OP claims to have played numerous matches before coming to his conclusion of performing worse during off-peak than during peak, statistics has a value in this discussion, since coincidences get smoothed out the larger your sample size of matches.

Gokil.2543:

As I said, I’m not being hostile. Logically, your explanation does not make sense. If he is forced to play with lower-skilled players, there will be 5 lower skilled players on the other team, and 4 on his team. Considering he is higher skill than them, he should be able to win more than 50% of the matches. I don’t understand why that is so hard to get. It’s the typical ‘ELO hell’ remark that people just don’t seem to want to let go of.

You’re oversimplifying things. Give me any player in the world, however good, and you can easily make them lose 50+% of the time using compositions of middle-tier players.

You see this in other sports, like basketball. There are players who absolutely dwarf the contributions of average players, and even the contributions of great players. I’m talking about transcendent stars like Lebron James and Kevin Durant. Yet these sorts of players can and do have seasons with records far below 50%.

The reason is that even all-time great players don’t raise win probability by anything close to 50%. (It’s because of how normal distributions work: they aren’t linear. They drop off exponentially as you get away from the mean. Sure, you might have a one in a million player. But they do not play a million times better than average. You would need a linear distribution for that.)

Gokil.2543:

This is just statistics. Judging by the fact OP claims to have played numerous matches before coming to his conclusion of performing worse during off-peak than during peak, statistics has a value in this discussion, since coincidences get smoothed out the larger your sample size of matches.

You’re getting the statistics very wrong. There are three viable theories: my peak-hour win distribution is accurate, my off-peak distribution is accurate, or neither distribution is accurate. Without any further information, it is impossible to know which theory is the most likely. Yet you opt, by default, for the most insulting one.

That aside, you are not even disagreeing with me. Obviously, I feel that matchmaking is working during peak hours and being unfair during off-peak hours. You think the opposite is true. Either way, we are both saying matchmaking is getting things wrong.

There are other possible explanations.

OP, are you using the same build in peak hours and non-peak hours? Maybe your build is quite good at doing its job, not messing up, and generally being useful. However, possibly it doesn’t perform as well at tasks like quickly killing low-skilled players, or surviving vs. 2 low-skilled players for extended periods of time.

GW2 PvP is pretty difficult. Even between, say, 60% and 75% on the solo queue leaderboards, there is likely a large difference in skill. It could be that in primetime, the MMR range is tight enough that your build isn’t being asked to do anything it can’t do. But in off-hours, let’s say your game ranges from 75%-60%, and you are at 67%. Even a 7% difference in MMR could be forcing you to do something your build can’t do.

As Taym said, some builds are better at “carrying” than others. Or perhaps you’re just not used to needing to “carry” your team, and play the same way in peak and non-peak matches.

Think of it this way: maybe Lebron James is theoretically good enough to carry a sub-par team to 50% victory vs. other teams. But to do that, he’s going to have to handle the ball most of the time and take most of the shots. If he doesn’t, it’s not a problem with the “matchmaker.”

there is indeed something fishy about match making

i did have games sometimes where one team would consist of rank 50-60 players and other team had rank 5-12.. i don’t mind low rank players as long as they perform well but theywere indeed pretty nooby and mostly just stood doing nothing; those are fights at top 1000-90%… i would really love to know how they even ended up in those matches… it is not fun neither for enemy team nor for the low ranked players themself imo

Gokil.2543:

If he is forced to play with lower-skilled players, there will be 5 lower skilled players on the other team, and 4 on his team.

Gokil…

That’s what you have gotten wrong about matchmaking.

The matchmaking algorithm currently selects 10 people from the current queue who have the closest MMR.

In low population times, these MMRs may not actually be close at all.

Then, the matchmaker distributes the players to two teams while attempting to make the AVERAGE MMR of each team as equal as possible.

If his MMR is high, he needs to have lower MMRs on HIS team, to keep the average MMR close to the other team.

This means he is forced to play on weaker teams if is MMR is high.

zilcho.7624:

Gokil.2543:

As I said, I’m not being hostile. Logically, your explanation does not make sense. If he is forced to play with lower-skilled players, there will be 5 lower skilled players on the other team, and 4 on his team. Considering he is higher skill than them, he should be able to win more than 50% of the matches. I don’t understand why that is so hard to get. It’s the typical ‘ELO hell’ remark that people just don’t seem to want to let go of.

You’re oversimplifying things. Give me any player in the world, however good, and you can easily make them lose 50+% of the time using compositions of middle-tier players.

You see this in other sports, like basketball. There are players who absolutely dwarf the contributions of average players, and even the contributions of great players. I’m talking about transcendent stars like Lebron James and Kevin Durant. Yet these sorts of players can and do have seasons with records far below 50%.

The reason is that even all-time great players don’t raise win probability by anything close to 50%. (It’s because of how normal distributions work: they aren’t linear. They drop off exponentially as you get away from the mean. Sure, you might have a one in a million player. But they do not play a million times better than average. You would need a linear distribution for that.)

Gokil.2543:

This is just statistics. Judging by the fact OP claims to have played numerous matches before coming to his conclusion of performing worse during off-peak than during peak, statistics has a value in this discussion, since coincidences get smoothed out the larger your sample size of matches.

You’re getting the statistics very wrong. There are three viable theories: my peak-hour win distribution is accurate, my off-peak distribution is accurate, or neither distribution is accurate. Without any further information, it is impossible to know which theory is the most likely. Yet you opt, by default, for the most insulting one.

That aside, you are not even disagreeing with me. Obviously, I feel that matchmaking is working during peak hours and being unfair during off-peak hours. You think the opposite is true. Either way, we are both saying matchmaking is getting things wrong.

I’m sorry, your comparison is completely irrelevant, there is no scrambling of teams in basketball.

It seems you chose to believe the matchmaking system picks on you specifically, and does not in fact work according to the MMR system. I can’t argue with that.

We look at what defines the difference between peak and off-peak, to see what causes the difference in outcome: there are fewer people playing during off peak. How could fewer people playing correlate to fewer matches won? There are not enough people around your skill level so the average skill level progresses toward the mean. Your skill level is higher than the mean → you win more, lower than the mean → you lose more.

That or its coincidence, which you ruled out in your original post.

That’s all I have to say about this.

NevirSayDie.6235:

OP, are you using the same build in peak hours and non-peak hours?

I don’t mean to make light of your post, but the answer is yes. I do swap classes every once in a while, but it never seems to change the pattern I’m experiencing.

Right now, the best predictor of my future performance is the first five games I play. My win percentage for a sequence of games seems tightly correlated with the percentage over the first five games in the sequence.

edit – I’m going to start keeping track of this stuff, so I can go on more than just memory. I know Anet has all this data already, but it would be nice to see the real numbers myself.

zilcho,

Gokil doesn’t know how the matchmaking algorithm works. But that is understandable. It was discussed a month ago, but many people have not seen what the developers posted explaining it.

zilcho.7624:

NevirSayDie.6235:

OP, are you using the same build in peak hours and non-peak hours?

I don’t mean to make light of your post, but the answer is yes. I do swap classes every once in a while, but it never seems to change the pattern I’m experiencing.

Right now, the best predictor of my future performance is the first five games I play. My win percentage for a sequence of games seems tightly correlated with the percentage over the first five games in the sequence.

edit – I’m going to start keeping track of this stuff, so I can go on more than just memory. I know Anet has all this data already, but it would be nice to see the real numbers myself.

Right, I was just pointing out another variable. It is entirely conceivable that your playstyle or builds could be better suited to matches with tight MMR ranges than those with wider MMR ranges.

I’m not saying I don’t believe your memory, just that your own performance is a variable. I don’t even mean you might have good days and bad days, you might just perform better in some situations than others.

To add some circumstantial evidence to my argument, consider that your win % is overall very close to 50%, which is the “perfect” number that the matchmaker should aim for. Claiming that the matchmaker is seriously flawed—and yet that it accidentally maintains an outcome nearly perfect—is a dangerous argument to make. It is more likely that the internal variations are accounted for by some other variable, such as your own performance, and that the matchmaker constantly fixes those variables to correct your win % to almost exactly 50/50.

“Matchmaking Never Worked”

fixed title for you.

Archaon.6245:

“Matchmaking Never Worked”

fixed title for you.

Well, I think it does work during peak hours. A lot of games have their matchmaking unfairly criticized. Which is why I waited a week to post this. I wanted to make sure I wasn’t just getting unlucky and I wanted to make sure my impressions were accurate.

I’m still just going by memory, though. I am ready to withdraw my concerns if the actual data contradicts me. (I strongly doubt that it does, however.)

Ive experienced play at both low and high MMR during the last week. Had a huge losing streak with a truckload of 4v5s (3 from the same guy DCing even) that dropped me from rank 200 till in the %s. So i think i have a good idea what the difference is.

Soloqueue as you all now is for a big part about luck. But at higher MMR your possitive actions quite often leads to something. If you keep 3 busy at their homebase, your team will benefit from it in most cases.

At lower MMR your efforts might be in vain in a lot of cases. The difference in skill amongst lower MMred ppl is absurdly big at times. So big that even a top player cant carry. Ive seen it happen so often during my own games or games i watched on streams.

The last soloq game i played for example was a blowout 500-50 win while we where 4v5: http://i.imgur.com/iYF8r2D.jpg. Teams like that arent carryable.

Justin ODell.9517:

@zilcho.7624: Are you taking into consideration the rating reset that happened about a month ago?

1) It could be that your MMR was reset higher than it was before. This should settle quickly because your deviation and volatility were also reset.

2) Playing during off-peak hours, especially while your deviation/volatility were high, could have caused your MMR to settle somewhere higher or lower than it should be.

Of course, this is just speculation. I will keep an eye on your account and see if I learn more.

Keep your eye on the system. Getting paired against top 50 players and then having people on your team in the bottom 900 doesn’t make sense unless this is some super backwards math involved in your MMR system.

It’s most likely not enough playerbase, even the best matchmaking in the world cant make up for a low playerbase.

Baldric.6781:

It’s most likely not enough playerbase, even the best matchmaking in the world cant make up for a low playerbase.

If there’s such a large deviation in MMR, honestly, I would rather sit in queue and play hot joins then to constantly be paired with people with so much lower MMR and play against people with such higher MMR.

Separately, when are we going to be seeing the ratings of the players in the match. I feel that the omission of this information is intentional to hide a broken system even further.

What I’ve noticed is my losing streaks are consistent and personal skill aside almost always results in being on a poor team comp versus a good team comp. I remember being on a lot of 2 war 2 thief teams today in soloq. I’d imagine I will eventually win but, when I lose it is a losing streak.