Guild Wars 2

Hi,

I have played my main account in sPvP, especially Friday and Sunday in order to complete my 200 warrior wins. I was already in Diamond, won 35 matches and ended this further in Diamond after win/lose/win/lose/…. and a “normal” win rate a little over 50%.

With that done, I decided to play my other account (Eowin the Second.6807) for the whole Sunday. Won 35 matches in a single day. I don’t know what kind of win rate that is, but it’s much more than I ever got before and much more than my own skill should produce. Went straight from late ruby to legend in a day.

Today I decide to go back to my main account so I may get a second Legend. Guess what ? I am only losing match after match after match. Win rate 1%. I get teamed with people who play like in emerald. I was 12 pips away from Legendary, I am now 24 pips away from Legendary despite gaining 2 pips from a win from time to time. Again, this is way worse than any bad streak I may have had over the past 4 years.

I am the same player playing the same profession and the same build at the same level, only using a different account. At this point I can’t just blame RNG, the streaks are way beyond what RNG can produce.

It would be nice if a dev could look into my pvp history on both accounts , because at this point it may point to a bug in the matchmaking (the infamous mmr hell/heaven some people rumor about ?)

Thanks

Eowyn

edit : after 6-7 hours straight of losing matches, I decided to give up and go back to my other account. Instaly won 542-24. This may be anecdotal evidence for now …

Your experiences have been reproduced by many people, myself included. It’s just the way MMR works, if you have good MMR you get good teams vs bad, if you have bad MMR, you get bad teams vs good. The game punishes people severely for losing, this is by design there is no bug.

Sooloo.1364:

Your experiences have been reproduced by many people, myself included. It’s just the way MMR works, if you have good MMR you get good teams vs bad, if you have bad MMR, you get bad teams vs good. The game punishes people severely for losing, this is by design there is no bug.

In no way does this explain what the OP is talking about. There are more complex problems within the algorithm than just new accounts having boosted MMR. I suspect that there is a flat out BUG in the system that is punishing accounts with more match history or plain and simple favoritism within the server.

It’s one of the two.

Trevor Boyer.6524:

Sooloo.1364:

Your experiences have been reproduced by many people, myself included. It’s just the way MMR works, if you have good MMR you get good teams vs bad, if you have bad MMR, you get bad teams vs good. The game punishes people severely for losing, this is by design there is no bug.

In no way does this explain what the OP is talking about. There are more complex problems within the algorithm than just new accounts having boosted MMR. I suspect that there is a flat out BUG in the system that is punishing accounts with more match history or plain and simple favoritism within the server.

It’s one of the two.

When I did this experiment and brought this topic up at the start of S3 I was told by almost everyone that this is just how it’s meant to be. Now we believe it’s a bug?
My main is over 3 years old, PvP from the start. You’re suggesting old accounts are affected?

Yea our old accounts that have been pvping for over 3 years are full of bugs, and new accounts are clean, that’s why new accounts win and old accounts only lose
My only question is, where can I buy some debugger for my old account so I can win on that 1 again?

It’s either a bug or something is going on that arena-net isn’t talking about. I’ve been following these problems for years now before ranked pvp seasons. People begin noticing it during ranked when they cared more about winning or losing but I noticed this same problem very early. Here is the thread that I kept up for quite some time. Read through the entire thread if you want real insight on what is going on with the acclaimed “MMR Hell”. Pay attention to the final page where the devs recognize and identify that there was indeed something wrong with the system, a bug. -> https://forum-en.gw2archive.eu/forum/game/pvp/High-MMR-is-punished-for-solo-que/first

After almost a year of following this thread and keeping on it, this was eventually posted by the devs -> https://forum-en.gw2archive.eu/forum/game/pvp/Updates-to-Matchmaking/first#post5524291

Point being: MMR Hell and some accounts being favored or unfavored within server settings has always existed and after seeing what players have been reporting during season 3, I suspect that something very similar is going on again. These people who are reporting 40+ lose streaks are not joking, they are not bad players and they are not lying. Something in the algorithm is causing certain accounts with certain numerics of match history and MMR to get rolled on the bad team almost every time they que.

What exactly is the problem? I have no idea because I am not a designer but there is quite clearly a problem. Players nowadays aren’t even complaining when they post! They are simply reporting that main accounts struggle much harder than newer accounts. This is not fair for a ranked spvp season and it needs to be fixed.

Srsly give me some debugger, I’ll pay 1,000 gems for it. (might as well spend my gems on something useful since they are obviously not ever gonna make any outfits sexy or beautiful enough to be worth buying)

I don’t understand why anet doesn’t do full mmr resets between seasons.

It’s not that the account is bugged. I suspect from the evidence of the effects, that the cause of the problem is as this:

If there is an old account with 8,000 – 10,000 matches played, the system identifies it as “an older better player” and expects that account to be able to carry. So the system uses that account with big long match history as a crutch to try and support weaker teammates within the division pip range “well it attempts to at least” but the system’s programming, it’s numeric values, are not realizing that it is impractical to expect one player to carry that hard. Carrying in GW2 conquest has limitations.

But it is more complex still. I’m sure most of us have noticed that the system has 3 rhythms that it places us in:

• Win Streak, this could go on for 10 matches or more sometimes
• That 50/50 win/lose/win/lose/win/lose, sometimes goes on for weeks
• Lose Streak, goes on for 10 matches or more, sometimes 40+ for those accounts that seem to be incurring the problem at hand.

We know that the system attempts to 50/50 your win/lose rate. This much is apparent. When you are under 50%, it puts you in to a win streak with easier matches. When you are over 50% it starts giving you progressively worse matches. When you hover at 50%, it likes to keep you right there with the win/lose/win/lose and it can go on for weeks before it breaks.

Important conclusion:
So the system seems to work exactly the way it should, when I hear stories from accounts that hover around 2,000 – 4,000 or sometimes even 6,000ish matches played as these numeric don’t seem to tag favoritism or unfavoritism within the algorithm concerning placement. The problem seems to occur in 2x situations:

• The player on the account with 10,000 matches hits a point where he is 50/50 for the season and the system goes in to the rhythm to attempt to keep him there but since the algorithm identifies 10,000 matches as some increased ability to carry, it will place him with weaker teammates and expect him to carry, even though the match is supposed to be fairly balanced. Where as the account with 2,000 matches played, will have the algorithm wait to find teammates who are better, even though both of these players are roughly the same actual skill level. -> matches played is not relative to skill level and should not be considered as MMR increase within the algorithm Arena net needs to remove this. As of now, total matches played is a huge factor in your team placement. It’s in the patch notes.
• Or that same player with10,000 matches of experience logs in to an alt account with only 400 matches played. Now he benefits the server identifying him as a player that “need a bit of support” even though he has identical skill factor. Regardless of what rhythm the server sets him on, his matches will always be a bit more favorable than the account that has 10,000 matches. Again, it is apparent and evident that this is happening. Read what people are posting this season and put 2 & 2 together.

The big thing to point out here, is that total matches played should have nothing to do with match making. I think that including total matches played in to match making during season 3 was a huge mistake and more than likely the source of the problem. Though, I wouldn’t put it past there being another bug.

It needs to be looked in to.

Trevor Boyer.6524:

We know that the system attempts to 50/50 your win/lose rate. This much is apparent. When you are under 50%, it puts you in to a win streak with easier matches. When you are over 50% it starts giving you progressively worse matches. When you hover at 50%, it likes to keep you right there with the win/lose/win/lose and it can go on for weeks before it breaks.

Basically, if you’re on a streak, spam games to get as high as you can before the system realises you’re on a win streak and “adjusts accordingly”.

Ubik.8315:

I don’t understand why anet doesn’t do full mmr resets between seasons.

Because that would make sense.

The system doesn’t have (at least, doesn’t need) any sort of memory to produce results like this.

Remember, the matchmaker does not try and produce balanced match-ups. It tries to match people within their pip bracket first and foremost; it then tries to get people with similar MMR, but stacks the best players of that group on one team.

This means that if you are one of the highest MMR players in your pip bracket queuing at any given time, you will repeatedly be put on the stacked team, creating an elevator to the top of the ladder. A side effect of this is that all the winning also pushes your MMR to the top, in a flywheel that keeps giving you better teammates to facilitate winning.

Weaknesses of the prediction / MMR algorithm in handling unbalanced match-ups (which I presume are there, because getting it right is really, really hard and there’s no way A.Net is throwing the manpower at solving it) exacerbate this effect; basically, you probably have instances where the MMR equation they are using thinks a team has a 20% chance of losing, but may only have a 2% chance of losing, because you’re on the edge of the range and the arithmetic / normality assumptions baked into the model don’t hold there.

So, using your case as an example – let’s assume you’re a well above average player and that correlates to a MMR of about 2000, and both of your accounts, at the beginning, have perfectly accurate MMR. Of your accounts, one in high diamond where 2000 MMR is typical (your main), and one in low ruby where the average MMR is, well, average (your smurf).

Your main is in high diamond. The matchmaker looks for matches for you, and finds other similar players with a 2000 MMR. Sometimes you end up on the favored team, sometimes the disfavored team, because there’s enough of a population to find players above and below you on the MMR scale. You get something akin to a 50% win rate at this level, and feel like you have stalled.

Your smurf is in low ruby. When you try to find a match, the matchmaker cannot find many other players with your MMR in your pip range, so it waits a few minutes longer to find a match. Once it does, since you are in the high end for your MMR, you are almost certainly going to be stacked onto the strong team, and get fed a bunch of above average but not on your level opponents. This happens throughout ruby, and by the time you hit diamond you have a MMR of 2200 from all the winning. This is high for low diamond, so you keep getting fed weaker opponents. By the time you’re even with your main account, your MMR is 2300, one of the highest in the bracket; you get paired with other players who are legit 2300 players playing on their mains, fed true 2000 MMR opponents, and get carried all the way to Legendary.

You go back and play on your main after that, get fed to players on a similar pip elevator a couple times; your MMR drops, you end up on the losing side a few times more often than expected, start to tilt, and drop a bunch of pips.

This will even out in the long run, but the long run here is 100+ games, while it only takes 30 or so to hit legendary. Timing of when you play is very, very important to hitting legendary, and winning in general – because, again, the matchmaker is not creating balanced matches, but unbalanced ones, and having a high MMR for your bracket means getting fed a steady diet of weaker opponents.

Still think people are giving matchmaking’s algorithm too much credit. It doesn’t look at how many games you’ve played, how well you’ve done on a class, or in a team comp, limit win streaks, or viewed which classes you were teamed with previously, or against, etc etc.

You’re just a number. Depending on how high or low it is will depend on which pool of players you’ll be picked from. Either you can make a big enough difference on the team to win games to force change mmr & pip placement, or you wont.

Your mmr on both accounts are just very different, amongst two very different player pools.

I don’t know how much it has to do with age of account. I went through ruby and diamond without a single loss on my main and only account, which has been there since beta. That still shouldn’t happen though, something is definitely wrong with the matchmaking algorithms this season. It seems to highly favor people with high MMR, my guess is so that it can quickly “boost” them to their “deserved” division. The problem that it is a bit too much and that it doesn’t seem to stop even once you reach legend, causing tons of bad games. Once competitive players reach legend, they want good matches so they can improve their skills, not stomps/getting stomped.

Anecdotal evidence here but my acct isn’t quite beta but very close maybe even day of launch and I won about 60 matches this season out of around 68…… so I don’t think it’s old accts are bugged. In my opinion it’s because I haven’t stopped improving. I didn’t tank my mmr on purpose like everyone else did season 1 and 2.

Yeah NeXeD I do hear that as well from time to time and I don’t understand it. You can watch pro player streams “guys with the highest MMRs in the game” and they still get matches that are impossible to carry and lose from time to time so how in the hell do some people win like that 60/68 matches? System is never putting these players in to the “targeted for elimination phase” and there is a lot more going on here than someone just having that high of MMR.

Look at some of these screenshots from when I approached the end of diamond. I started taking them to show how no matter if I won or lost, I was being forced to heavily carry matches, all of them. In the end, I received more matches that were impossible to carry than ones that could be carried. This went on for 20 matches or more and it’s a miracle that I won any of these matches at all if you had seen how these players were playing. My big question is: If the system’s match making works the way arena net says it does and the way people are led to believe it does, why does this series of screen shots look like my match making is a little out of place?

This is the MMR Hell that people talk about. It’s real.

Trevor Boyer.6524:

Yeah NeXeD I do hear that as well from time to time and I don’t understand it. You can watch pro player streams “guys with the highest MMRs in the game” and they still get matches that are impossible to carry and lose from time to time so how in the hell do some people win like that 60/68 matches? System is never putting these players in to the “targeted for elimination phase” and there is a lot more going on here than someone just having that high of MMR.

Look at some of these screenshots from when I approached the end of diamond. I started taking them to show how no matter if I won or lost, I was being forced to heavily carry matches, all of them. In the end, I received more matches that were impossible to carry than ones that could be carried. This went on for 20 matches or more and it’s a miracle that I won any of these matches at all if you had seen how these players were playing. My big question is: If the system’s match making works the way arena net says it does and the way people are led to believe it does, why does this series of screen shots look like my match making is a little out of place?

This is the MMR Hell that people talk about. It’s real.

If i took screenshots of my 100+ games 50/50 win lose, they would appear exactly like your ones. I have been forced to carry pvers or new hot players during all the seasons, with every game no matter if lose or win always been at top scoring, top kills, top dmg dealt, top skirmish etc… with my teammates rallybotting left and right not even cleansing one single time, and this is coming from someone who, differently than you, didnt played a self sustained class, i played revenant all the games, a class that rely heavily on teammates support/cleanser in order to win team fights.
This carry kitten has happening only because of my 7k+ pvp matches
Unfortunately, this is a 5v5 game type, and no matter how good i can be, i won’t be able to carry games in 1v5, especialy with the heavy counters mechanic that exist in this game.

oh also i can add i was forced to run retribution instead of invocation,wich resulted in a better win % thx to stability, because not only i had to kill every opponent, but i also had to stomp them very quickly before my teammates start rallybotting them!!!

The game seems to punish those with a relatively high MMR that did not advance fast enough through to legendary in the first 3-4 weeks.

That has been my experience through all 3 seasons.

Through grinding even super really terribad players can get in to Ruby, unfortunately when your pip range is saturated with new players those are the types of matches you are going to have.

Zoricus may be right actually. During S1 I only played about 3-6 matches a day the entire way through. During S2 I started play about 4 weeks late due to a large move. During S3 I played lightly 1-2 games a day for the first couple weeks and then had to take a 3 week break middle season due to an eye injury. It does seem early season in all the season, I had better matches. Match quality kept on dropping the longer I waited to try and grind to legend at the kitten end of each season.

Trevor Boyer.6524:

Zoricus may be right actually. During S1 I only played about 3-6 matches a day the entire way through. During S2 I started play about 4 weeks late due to a large move. During S3 I played lightly 1-2 games a day for the first couple weeks and then had to take a 3 week break middle season due to an eye injury. It does seem early season in all the season, I had better matches. Match quality kept on dropping the longer I waited to try and grind to legend at the kitten end of each season.

It’s because all the decent players powered through the divisions and quit. Thus by the end of the season it’s just the grinders who took 500 games to get to ruby and the late starters or “slowpokes” left (with the occasional smurf).

Ubik.8315:

I don’t understand why anet doesn’t do full mmr resets between seasons.

It takes more than 2 months to get stabilized.

Look for volatility.

Trevor Boyer.6524:

• The player on the account with 10,000 matches hits a point where he is 50/50 for the season and the system goes in to the rhythm to attempt to keep him there but since the algorithm identifies 10,000 matches as some increased ability to carry, it will place him with weaker teammates and expect him to carry, even though the match is supposed to be fairly balanced. Where as the account with 2,000 matches played, will have the algorithm wait to find teammates who are better, even though both of these players are roughly the same actual skill level. -> matches played is not relative to skill level and should not be considered as MMR increase within the algorithm Arena net needs to remove this. As of now, total matches played is a huge factor in your team placement. It’s in the patch notes.

Perhaps time played and skill is not relevant. But please, if you fix my account 10k games… make sure I am not paired with 100 or less games player.

I am all for placing veterans together to play versus each other. So, I don’t care having big lost streak because of that.