Guild Wars 2

Flyphish.6398:

If the vast majority (i.e. >80% just guessing) are as you say then what is so “nuclear” about going to 100%, or hard filters? Queues are really very fast. Does eliminating outliers really add to queue times?

My perspective being that a queue time of 5 – 10 min is perfectly fine, other opinions will vary.

For most players, the increase queue times will hardly be noticed. For the outliers, it will likely be several fold worse.

Lopez.7369:

Isn’t this why you guys added solo queue in the first place? I’m really scratching my head here.

We added it for the same reason we removed it. Player demand. We can make it work, even if we have to use the nuclear option of hard filters.

There are a few things going on, and we need to pick threw the data to see if we can find the truth.

Based on the data alone so far, I’d say we’re at least 90% toward where we all want to be. Just need to drive home that last bit. The vast majority of matches are solos vs solos. When they aren’t, the groups are usually spread on both sides pretty evenly.

When its wrong though, it tends to be really noticeable. It’s this last bit I’m attacking next.

Lopez.7369:

I just don’t like how many groups, especially full premades, I face with solo players. If that was fixed, this thread wouldn’t exist.

Right now, I am facing a team with a two-man and a three-man. I have five solo players on my team.

Is it specifically that you don’t like facing them, or that you don’t like feeling overwhelmed and losing to them? We can fight that problem from multiple angles. My preference is to avoid hard filters because they increase queue times and only bypass the issue rather than address it.

Tellerion.8102:

Matchmaking doesn’t seem improved because it’s simply not.

Come on guys. I’m still working on improving it based on real data. Help me out and give me specifics. Don’t make generalized statements like this unless you’ve got something to back them up.

We can improve things, we just need to work together.

GhOst.4019:

A grace period would be good…60 seconds? Also, for the whole team getting deserter results, that means -3 pts automatic on LB’s right?

I don’t know what the grace period would be. Feel free to debate the pro’s and con’s of different lengths.

GhOst.4019:

Also, for the whole team getting deserter results, that means -3 pts automatic on LB’s right?

Correct. The alternative is to allow parties to dodge bad ladder results by having just one person take the fall. Another alternative is that we just do nothing, and everyone that suffers a 4v5 takes a loss (unless of course they manage to still win).

Bugabuga.9721:

Wooow. Oooohkay Does this imply intent of “do not ever queue as a party. EVER”? Because if someone disconnects you get whacked with deserter result?

This change is not live. This change would likely have a disconnect grace period. Know someone that DCs a lot, and for a long time… probably not a good idea to party with them.

Been slacking, needs more mesmer…

These notes were added by mistake. They’re upcoming changes we’re testing, but I can’t say if and when they’ll make it your way.

The intent was that buys would update ladder standing the same as a win, but would not update matchmaking rating. Forfeits would not update ladder standing (for the loser) or matchmaking rating for either team.

The player that left would get a desertion on there record, and will take a full loss whether their team won or loss, but would not affect their matchmaking rating. A win/loss only changes to a buy/forfeit if the losing team had desertions.

To prevent exploits, all members of the same party as the deserter would also receive a desertion result, but will not receive dishonor unless they leave early as well. Due to their strict nature, these changes would likely be accompanied with a disconnect grace period.

Does that answer your questions?

Becka Williams.4978:

Wasn’t this what the whole pvp revamp was supposed to do? Why do we have to wait for another revamp to fix the problems with this revamp?

We’re like to iterate and continue to improve. Something we can do much faster with the back-end servers.

Silentshoes.1805:

Yup.
And the mean is 1293. Median is lower.

http://www.alcula.com/calculators/statistics/median/

Yeah, I don’t know where my head was with that one. Probably already halfway home.

insanemaniac.2456:

Justin ODell.9517:

Silentshoes.1805:

Would median MMR make more sense?

No, not necessarily. E.g. 1200, 1215, 1210, 1240, 1600, the mean is 1293, but the median is 1400.

um that median is 1215 :X

*whistles and walks away slowly*

lilwizz.5783:

I mean you said it yourself, MMR is more accurate, then why not make the Leaderboard just based on the MMR?? The system is way too inaccurate.

The beta ladder is an experiment for one piece of a larger whole. The reason we call it beta is because it isn’t the final product.

Henry.5908:

I means in this patch, anyone who not in the game when it ends.
It doesen’t register you so no win no lost?
You just minus point becuase system think you are dc or change another char?

If your aren’t in the game when it starts, or leave at any point, yes you will get flagged as a deserter and lose the max amount of points for the match.

We’ve got something coming down the pipe that should improve these sort of scenarios. Details soon.

Silentshoes.1805:

Would median MMR make more sense?

No, not necessarily. E.g. 1200, 1215, 1210, 1240, 1600, the mean is 1293, but the median is 1400.

Henry.5908:

when you know you will lose the game,then you leave the game you just minus 3 point and it didn’t record any lose game ?Is this a Bug?

Losing 3 means either your team score < 200 points when you were expected to win, or because you left the game before it completed.

Thanks adozu! Here are the links to go along with what you were referring to.

adozu.6398:

this is from an official anet source

http://wiki.guildwars2.com/wiki/PvP_Matchmaking_Algorithm#Ladder

adozu.6398:

even tho they don’t articualte on how the winning expectation is calculated exactly.

http://wiki.guildwars2.com/wiki/PvP_Matchmaking_Algorithm#Match_Prediction

Crell.6401:

Are people recieving dishonor if they’ve tried to leave the queue and decline (or are offline) when it pops? If so, can we get a reset after this gets fixed? =)

Dishonor is currently disabled.

Silentshoes.1805:

“One of the biggest problems is that we simply aren’t messaging any of the details to you in-game”

Ummm.

Like your odds of victory, what score you have to get to get N points, etc..

Yep. Verified that it was broken last night, and fixed it last night. Fix is going through the pipeline and will show up for you guys soon. Unless, that is, we find something wrong and have to hold it back, of course.

Thoth Divine.8642:

queue times for top players have shortened in compromise for less even matches, and, thus, teams are more often facing solo players or small teams with a few pugs.

This has always been true. It happens less with the new matchmaking because we start this process later.

Thoth Divine.8642:

It was supposed to improve play by reducing the amount of times disliked maps would get chosen

It has. Our least popular map (I won’t say which) has gotten 4% of votes, and has only been selected in 4% of all games in the last 24 hours. It use to be uniform, which meant all 6 maps had a 16.7% chance of being chosen.

Thoth Divine.8642:

Last but not least: Leaderboards.

It’s a test scoring system, no the final one. One of the biggest problems is that we simply aren’t messaging any of the details to you in-game. That said, we will be taking all feedback into consideration.

Mrbig.8019:

When i duo/trio q with lower or slightly lower ranked players, match ups seems to be WAAAAAY better, prolly cuz it averages our MMR.

Very likely, I’ve heard this from other players as well.

butch.8136:

2 minutes minimum, 8 minutes maximum

2 minutes average. Some people will be higher, some lower.

altqq.5481:

people matched with +10/-10 of their own rank, wouldnt that be the whole point of the matchmaking system?…

No, matchmaking focuses primarily on MMR, which tends to be far more accurate than rank. Of course we can do more to reduce the rank gap, so I’ll take a look at your games and see if anything stands out,

Mrbig.8019:

It’s not like i don’t want to face them, it’s like i don’t want to face them if they’re FULL TEAM while i’m SOLO with other SOLOERS.

I get you. We’re collecting data for this entire ladder, and don’t think the ladder scoring rules are the only things we care about.

We’ve been discussing the issue and working on changes that should help with this. The only thing likely off the table (for now) is hard-filters.

Mrbig.8019:

In order to stop esl teams to go vs me ?

Hmm, your MMR is pretty high. Looking at your stats, it doesn’t seem odd though.

Is the issue that you’re getting matched with these players while queuing solo? Or is it that you don’t feel you should be matched with them at all?

Do you have a specific match you’d like me to look at? If so, do you remember the time?

Vamp Rook.7835:

Add a minimum of r20 and 50 unranked games for play ranked games. It’s easy.

So the issue for you is getting matched with inexperienced players? I don’t want to toy with the matchmaking rules for ranked while the ladder is active, but I’ll talk with the team about possibly increasing the PvP rank distance scoring for unranked arena.

We’re also looking into long term solutions to the problem, so thanks for the feedback.

Do you have a specific concern that I could address?

masskillerxploit.2165:

like, there isnt enough soloq players to match up? at a given time or what.. this is severely annoying.

Out of your 26 matches in the last 24 hours, and excluding the ones where you were with a party, all but two of your matches have been all solos. You even won one of those.

Are you sure it’s the pre-mades that are bothering you? Is there something else we could do with matchmaking to improve the quality?

Edit: Here are your wins vs losses based on party size.
1: 6 — wins: 4, loss: 2
2: 4 -- wins: 2, loss: 2
3: 1 — wins: 0, loss: 1
4: 1 -- wins: 1, loss: 0
5: 14 — wins: 9, loss: 5

I don’t see anything unreasonable there, do you?

DaveGan.8065:

so can you at least try to put them on different teams?

It does try, in fact the scoring on checks for duplicates at the team level. It’s just being overshadowed by other things.

Part of the problem is that mm is holding these players back, trying to find them matches where they’re unique… but eventually it gets overshadowed by other scoring rules (like the age score, which is used to compensate for long queue times) and they all dump into the same match.

Another part of the problem is we will not do anything about pre-mades with duplicate professions, other than the scoring penalty they take, but that only means longer queue times.

The scoring that works against multiple of the same class is working… there just are so many it can’t hold back the flood.

Chaith.8256:

It’s simply because you lose points for DCing or character swapping.

The data makes no distinctions for the cause of a desertion, but yes this explains the discrepancy on the leaderboard. A future update will include desertions in the losses column.

Chaith.8256:

It won’t affect your win rate unless you don’t return, just leech the points.

Coming back won’t matter. You are awarded dishonor and flagged as a deserter the moment you disconnect. You’ll still get end-game rewards if you come back though.

Lupanic.6502:

if this is correct why Ostricheggs has 14 games won 1 lost and he has 9 points?

lets asume for every win +1 makes 14 points. For 1 loss -3 is 11 not 9

Is it because Ostricheggs is esports?

Hmm, that is interesting. I’ll look into this.

Zietlogik.6208:

Didn’t they say that Skyhammer, Spiritwatch and Courtyard aren;t in the rotation since the patch? I have yet to see them…

Yes, they are no longer available in ranked arena. However, I didn’t make that change until an hour after the ladder started. -_-

Mrbig.8019:

You always get at least 1 point for winning, and lose 1 point for a loss.

Sorry if someone already mentioned this. This is incorrect. You will always get at least 1 point for a win, but you will not always lose 1 point for a loss.

The full range for a win is +1 to +3. The full range for a loss is -3 to +2.

Laserbolt.6731:

I find it so interesting to see how the new system is grading people’s skill.

These people are consistently winning matches that their team is expected to lose, I guess. Mostly they have wins.

None of the typical names appear in the top 10, unless I am missing someone well-known.

It looks like you cannot climb the leaderboard by crushing people you are expected to crush.

Maybe some of the top players would do better playing solo and carrying weaker teams.

Based on those scores, I wouldn’t say they were not expected to lose, merely that it was not in their favor.

Will be interesting to dig into the data and see how it’s playing out over time.

If this is an issue of their MMR not matching their skill, it should resolve itself after a while.

Kharr.5746:

So, I just finished a ranked arena match (started ~ 5:00 pm EST) and it STARTED as 4v5. We did not have a 5th from the very beginning, and no one joined all game. I thought the new system was supposed to prevent these matches from starting? Why are we still starting with 4v5 matches??? Someone obviously didn’t click the queue to join, yet the match started anyway.

The match cannot start without a player confirming the match. There may be some race condition if the player leaves after accepting caused by not receiving the message about the player leaving until the match is already started. I’ll look into it and see if there is anything more I can do to prevent it.

I found an idempotence failure within the dishonor code.

Dishonor is disabled until the issue is resolved.

Found an idempotence failure for dishonor. Disabling it until it’s resolved.

Groutesk.6082:

bun can you explain how it stacks?

http://wiki.guildwars2.com/wiki/PvP_Matchmaking_Algorithm#Dishonor

savov.3712:

Thank you for the quick response.
I’m playing only solo ranked, since the leaderboard started, like 3 hours ago(?). All my games from today, are all solo ranked and all are since the leaderboard started. Haven’t played anything else. And I’m yet to be placed in a fun, fair and competitive match. Till now, all feel like playing low-level hotjoin…

Not sure what to tell you.

The lowest rank you played against with today was a rank 30. The majority have been 70-80. You haven’t had any roster size mismatch issues, and MMRs have all been in range for each match.

You only receive ~50 minutes of dishonor if you have 20 stacks of dishonor. That’s two abandons.

I’ll still take a look your account just in case.

I’ll look into it.

In the future, all I need to know is which arena you were in, and what the time it was.

Please include the error codes if you get them.

Fixed. Sorry!

/15charr warriors

Zanthrax.6538:

I assume the test season hasn’t officially been turned on yet? Phanta on stream just got courtyard pop for ranked.

Oops, my bad. They’re removed now.

You will now get dishonor if you queue dodge or abandon a game.

Queue Dodge: 3 stacks, 6 minutes for the first offense
Abandon: 10 stacks, 17 minutes for first offense

The ladder will be begin collecting data in one hour.

PST: 11:00
UTC: 19:00

Good luck everyone!

Spectremercy.5904:

So I am thinking about when I was in a ranked game, I was playing but somehow a que pops and I just clicked “yes” then I get to choose a map and go to the new map.
Does this considered as desertions or bugs? I would not click “yes” if I know my points would go down.

This shouldn’t be happening anymore. Are you sure that is the case? Either way I’ll look into it some more.

If you do get a queue pop while in a match, hit cancel. This will not affect your ladder standing, but leaving your current match will.

insanemaniac.2456:

dont cancel queue pops

This will not affect your ladder standing, but will give 3 stacks of dishonor when it’s enabled. If you had no current stacks, that’s about a 6 minute timeout.