Guild Wars 2

Rising Dusk.2408:

• I’d rather wait longer and have fewer games overall if those games are of higher quality. If I get stomped 100-500 I stop playing PvP that day because it makes me angry, as I’m sure it does for a large number of others.

how often does this happen tho? it only happened to me at times when it was reset, after like a dozen games or so it happens every once in a great while (at least for me).

• I’d be totally okay with really long queue times if we could bloody leave HotM or have something to do. An 8 minute queue is only bad because we’re forced to stay in HotM and there is nothing to bloody do in there. There aren’t even crafting tables or a Mystic Forge, let alone a jump puzzle or anything to keep us occupied.

this solves the whole queue length issue quite nicely. sadly i doubt it will be fixed any time soon because of pride. not wanting to admit that an idea was stupid and pushing it on people is never a good solution. i really dont understand what is so hard about adding the crafting elements to the game while in the mists. heck add in some gathering nodes too.

Justin ODell.9517:

Cindas.6149:

Second, if “being matched against a premade when you’re solo is rare”, why have it be a possibility at all? If it is rare, what does it give to the players and the system? It appears it is only reducing the queue time for “high skilled players”.

We also allow it because a lot of the time it doesn’t matter, especially now that we inflate party MMR by the number of members to compensate for the advantage.

Math: party.rating *= config.party.inflation * pow(party.size, config.party.curve)

5 players with a mean of 1500 becomes 1940 with inflation at 5% and the curve set to 1.1.

There is still that occasional match though that gets totally ruined by the fail-safes. I’m not certain waiting longer for those players would improve quality for them or just reduce the number of games they get to play.

See people, told ya something like this happens. (premades get boosted points, so in theory a weak premade would face medicore players).
Only times you should lose to premades (like harcore premades) is when they are the only team queing and there que timer reaches that limit where it forces them to face whoever is qued.

Solo players in this game are able to beat premades, most of those premades are nothing anyways. I’d rather get teamed with some of the good solo players I’m playing with and face a premade than be que’d with newbs and have a who carries harder fight.

Me and another good player with 3 newbs vs 3 good players and 2 newbs is what I usually get.
I’d rather have a me and 4 good players vs a premade.

There system of detecting good players is bad, they need a player ranking system to tell whose good and not, so whenever I do go against premade its not me and 2 good players and 2 newbs vs premade (hence we lose, 2 weak links).

Remember when I was talking about beating abjured, how those teams facing abjured have weak links? Well I’d assume the system gives me a few weak links when I face a premade.

Hate playing with weak links. If you play with me, I communicate via map pings, and if I’m pinging the map hard, that means I’m angry. I ping on you and ping the place your suppose to go and you go do something dumb, ya you the weak link. I have good map awareness, I know most of the time where people should be at.

oneply.9586:

how often does this happen tho? it only happened to me at times when it was reset, after like a dozen games or so it happens every once in a great while (at least for me).

It actually happens to me fairly frequently. I basically don’t have time to play PvP at the moment except during Oceanic prime time, which results in 5-8 minute long queues with total blowouts either in my favor or against. I haven’t PvP’d in a few days because it’s just been so frustrating, but I posted my most recent one in that other thread where Justin requested information on it.

uberkingkong.8041:

There system of detecting good players is bad, they need a player ranking system to tell whose good and not, so whenever I do go against premade its not me and 2 good players and 2 newbs vs premade (hence we lose, 2 weak links).

and thats the big problem right there.
i have a feeling points farmers get marked by the system as good/average players, when really they are bad and cause their team to have to overcome their selfishness to win the match.

Saiyan.1704:

Shinzan.2908:

So what’s the alternative you complainers want? Matchmaking just keeps looking for a match until the right conditions are met even if that would take 10-30 minutes or even longer?

Premade vs solo is never a good condition.

You didn’t answer the question. Again. If mathmaking can’t find a good match, should the team not get to play until it does?

Justin ODell.9517:

uberkingkong.8041:

Me and another good player with 3 newbs vs 3 good players and 2 newbs is what I usually get.
I’d rather have a me and 4 good players vs a premade.

It works on both sides. The players added to your team will be at the inflated rating. I.e. if your regular rating is 1500 but after party size inflation it becomes 1700, the missing spots will be populated with players as close to 1700 as possible.

uberkingkong.8041:

There system of detecting good players is bad, they need a player ranking system to tell whose good and not, so whenever I do go against premade its not me and 2 good players and 2 newbs vs premade (hence we lose, 2 weak links).

Not sure your logic follows given the misunderstanding above. To put it as simply as possible in this context, the rating system doesn’t care who is good and who is bad, it only cares who wins and against whom they were playing.

oneply.9586:

i have a feeling points farmers get marked by the system as good/average players, when really they are bad and cause their team to have to overcome their selfishness to win the match.

By point in “point farmers” do you mean ladder points or individual points? Ladder points work off of your team’s performance, so there should be no conflict of interests there.

That’s where your matchmaking problem is.
A newb is a newb.
You could have 2 1600s and 3 1200s vs 3 1500s and 2 1100s.
The 3 1500s going to win because they know how to play the game and aren’t dumb.
Look at my idiot type mistakes vs common mistakes wherever I posted that.
newbs are likely to make idiot mistakes, simple rating isn’t going to help. But yes if that’s all you have queing than its fine. But if you have more, its better to make it fair, 2 pros 3 newbs vs same thing.

weak links are very crucial to the game. especially when its tiny game 5v5. I cant imagine a 2v2 3v3 with 1 weak link. That’s 50% or 34% of my chances right there.

Don’t listen to those 2v2 3v3 folks, if anything make this game 8v8 so weak links mean less.

Also, your we rate people on wins. Like many have said, people get carried. That’s a stupid way to rate people. You need to use a player rating system.

Because if you do a 2 strongs 3 weaks vs 3 strongs 2 weaks. However, the catch is that 2 strongs, 1 guy is actually a weak, that’s 1 strong 4 weaks vs 3 stongs 2 weaks. That’s a blowout game.

You seen people post about blowout games, that’s why they happen. You need to start working on the player rating system. Not only does it benefit matchmaking, it gives people a way to judge themselves because you also hear people talk about “how can I tell if I’m doing good or bad”, well now ya have a way. Also, creates competitiveness with friends and rivals. You just cant go wrong, start working on that kitten now.
And yes the player rated system should be public. Everyone should be able to see if your a newb or if your a pro. There should be stars by your name too indicating it, something nice to have and show off. “hey buddies, see I’m a 5 star, I’m so awesome”
2 things, skill rating and popularity/how you act/do people like you type rating.
Because people will give you bad ratings if they dislike you, but if theres 2 separate ones, more likely to show truth in the rating besides someone hating on you hence you get a 1 star even though they hate on you because you sat on them by owning them so you should be 5 star skills 1 star popularity.

Justin ODell.9517:

oneply.9586:

i have a feeling points farmers get marked by the system as good/average players, when really they are bad and cause their team to have to overcome their selfishness to win the match.

By point in “point farmers” do you mean ladder points or individual points? Ladder points work off of your team’s performance, so there should be no conflict of interests there.

no not ladderboards, rank/reward points

Justin ODell.9517:

We also allow it because a lot of the time it doesn’t matter, especially now that we inflate party MMR by the number of members to compensate for the advantage.

Math: party.rating *= config.party.inflation * pow(party.size – 1, config.party.curve)

5 players with a mean of 1500 becomes 1845 with inflation at 5% and the curve set to 1.1.

I’m not sure I understand this correctly. Does it go like that:

rating = rating * inflation * (party_size-1)^curve = 1500 * 0.05 * 4^1.1 = 344

Shouldn’t it be:

party.rating += party.rating * config.party.inflation * pow(party.size – 1, config.party.curve)

This would give 1844 party rating.

The problem with player ratings system and glicko is there is zero chance of ever arriving at a statistical model that accurately measures a person’s worth in a team game.

The unobserved data isn’t like models in econometrics. A lot of unobserved data in matches like little things that turn team fights around need to be accounted for to ever make a truly make a player ratings model work.

like what? boons applied to other players? combos participated in (both fields and finishers)? healing others? i’m pretty sure that people could find a way to to focus only on boosting their own personal ratings in those categories to the detriment of the team. can you imagine finding someone so focused on it that they’d lead the leaderboards in those stats categories, but yet have an overall 75% losing record? (ie they lose 75% of the matches they play because they’re too focused on making their stats look good instead of actually helping their team win)

SobeSoul.6910:

The problem with player ratings system and glicko is there is zero chance of ever arriving at a statistical model that accurately measures a person’s worth in a team game.

The unobserved data isn’t like models in econometrics. A lot of unobserved data in matches like little things that turn team fights around need to be accounted for to ever make a truly make a player ratings model work.

other games with ratings systems are perfectly legit and accepted with only needing to know win or lose and who you played against

Justin ODell.9517:

SobeSoul.6910:

The problem with player ratings system and glicko is there is zero chance of ever arriving at a statistical model that accurately measures a person’s worth in a team game.

The unobserved data isn’t like models in econometrics. A lot of unobserved data in matches like little things that turn team fights around need to be accounted for to ever make a truly make a player ratings model work.

I have to disagree with this, but if you can come up with a better idea I’m interested.

Think formulating a model for winning would be the first logical step.

winning = B0 + B1kills + B2resses B3pointscaptured + B4pointsdefended + B5outgoingheals…. u (something of that nature)

Then hypothesis testing until we derived a formula correlated the best with the most successful people who queue solo. You could use the model to check how much other players deviated from it to figure out their rank. Obviously I have no knowledge to code any of that. However, I do know that plain wins and losses is not a solid method to evaluate a players’ ability to perform in game when there’s so many factors that go into winning and losing in this game.

As my econometrics professor told me, the Oakland Athletics used such methods to figure out the best bargain for productive hitting was.

To add to that post: I think if you could quantify what it takes to win through linear regression and modeling, it would also give you some ability to make charts on what players could work on to get better.

For everyone complaining about solo please read all the posts

simply put there is a system based on your win percentage against the teams you plays rank. Im sorry but im willing to say the majority of you asking for solo que arent that good. Im also willing to say that your principles are much different then my own. Im sorry if 4 other people qued up and there good and im put in that group and we have to fight the abjured, ok bring it on. Let me have this chance, you literally have people on here complaining about fighting any pre-mades. Seriously what? You have other people trolling justin, really?

PVP is competitive, in what sport is the game completely fair? Seriously the same 4 guys win tennis every tourney, basketball has been dominated by 5 teams in the last 15 years. There is thousands of smucks playing this game and for some reason you come on here asking for a hand out when you can see they have a great system in place and is trying to get better.

Last note, no i will not wait any longer for ques. Ive lately had alot of 5+ min ques whether i solo or partially team que. If it goes back to the old days when we had solo and team ques ill do something else and at this point it might be wait for the expansion since theres nothing left for me to really do.

kdaddy.5431:

PVP is competitive, in what sport is the game completely fair? Seriously the same 4 guys win tennis every tourney, basketball has been dominated by 5 teams in the last 15 years. There is thousands of smucks playing this game and for some reason you come on here asking for a hand out when you can see they have a great system in place and is trying to get better.

So what you are saying is that sPvP was designed for “smucks” in the last two years?

In light of it all the matchmaking has been fine for me as well. A recent game I was frustrated because I though I was playing with inexperienced players… on my team and the opposing team. It turned out being a close game; it was challenging in its own right.

Justin, keep my mmr to yourself please

Justin ODell.9517:

uberkingkong.8041:

Also, your we rate people on wins. Like many have said, people get carried. That’s a stupid way to rate people. You need to use a player rating system.

I believe TrueSkill uses player ranking in the game when updating ratings, but I’m not sure that would work as well for us. I’d love to use TrueSkill, but it has a patent and I hear Microsoft doesn’t license it out to competitors. Also, our point structure isn’t designed for this, and would probably need a bit of an overhaul.

People being carried only matters if they are constantly carried by the same people, but it is true that this causes their ratings to take longer to reach the same accuracy.

I’ve been toying around with my own rating system in my spare time, but it’s not well tested or verified, never mind production ready.

never even heard of true skill.
What I think of when ratings is like yelp.
Straight up legit feedback.
You should let people be able to comment on players too.

You should play a game, when its over it and after you get ported out it should allow you to rate the players on your team and opposing team. Shouldn’t be forced or maybe should, up to you.
2 ratings, skill rating and popularity
also have a place for people to comment on those players too.
So example be like 4 stars skills 5 stars popularity (comments this guy knows how to target, communicates too, hes awesome!) or a rating like 5 stars skills 1 star popularity (this guy very rude player, he likes to let you know he owned you, talks a lot of trash).

doesn’t have to be stars if you can use something like thumbs up, use like currency, this guy is a platinum player (5 stars equivalent) this guy is a bronze player (1 star equivalent). Do what you gotta do to get by the patent. Make it related to GW2 somehow.

You know I kinda think of this not only like yelp but as the bodybuilding.com ratings.
they use something called rep. if people in the red on that site they most likely rude or don’t know what they talking about, but if someone has like 10000 rep power, that guy knows what he talking about, he the kitten. They have other ratings too, like muscle rating or something.

All I say that would work is 2 different ratings skills and popularity and have comments. Make it feel kinda like social media. Gotta get people to interact and play together, not solo mindset.

Skill rating for better matchmaking, use your own system along with the skill rating. popularity rating and comments just for fun and some people will make that a grind or something. Yall need to stop making this game super serious, there needs to be fun.
Like on bodybuilding.com I think some people enjoy being super red, and others they like to get crazy amount of rep power. Good way to look at the community through popularity, some people don’t like the skilled players but other people are popular, like ostrich eggs, his engy video made him pretty popular, even though he’s a WTS winner. Some of those skilled players, aren’t very popular and its good thing for the players to compete with too. Like most popular streamer. He could suck but people like to watch him and think they cool so he’d have like a 2 star skills 5 star popularity.

Like SAB in PvE, that’s nothing but fun, a lot of people want that kitten back. The rewards wasn’t even that great compared to other grinds, but people like to have fun every now and then.

Haha, that would just lead to downvote type trolling.

SobeSoul.6910:

Haha, that would just lead to downvote type trolling.

Nah I see it being useful.
Good way to check to see whose the best x profession.

I bet cruuk, caed, jumper, toker would love to see whose the best by community votes.
As well as other professions like engy, darth, chaith, ostrich, whoever else.

Also, popularity like whose best streamer or best video maker (engy tutorial).

Like I would make it so you don’t have to put both skill and popularity, but chose. Like for example. Man I loved ostrich eggs nades video, gonna give that guy 5 stars and give him a comment thanks really helped me be win more!!!

If people downgrade people on skills that’s just dumb, it wont work. Because if people disliked cruuk and he’s playing with newbs, those newbs are going to give him 5 stars because they sick of getting owned by that guy, he shouldn’t be que’d with them.
That’s the point of it, if you sick of going against someone that’s just too good, give him high rating so he stops playing against you and plays against people that kitten him up instead of him kittening others up.
Samething with newbs, you hate that guy on your team but he keeps getting que’d with you because system thinks hes good, give that kitten a 1 star so he goes away and gets que’d with other newbs and not good players like yourself.

This gives you some say in the game, like a vote. Your vote matters.
1 star that newb, and comment, 1 star this kitten, he’s a newb.

This too many kids play this game is probably only thing that’ll ruin this. Lemme ask ya, whose the majority? How many kids do you really think play. How many of those kids do you think give a kitten? I mean I played GTA, that game was wayyyy popular, and it was totally anti “my kids play it, stop the nasty language”. Those kiddy friendly games don’t work. If anything, use the words kitten or —-- to block those bad words people leave for your comments.

I think of WoW as a kiddie game, yeah it might be big, but its reputation, that’s a nerdy game, kids play that game.
GTA, has a good reputation, this game is fun, positive stuff, cool people play this game.
You choose your reputation.
To be honest I think GW2 PvP mostly with starcraft. That game had a good reputation, wasn’t kiddy or anything. Nothing but competitive people. Competitive reputation.

And if people bad mouth you, you fight back, give that guy that bad mouth you via comment a 1 popularity. Also, maybe comment the comment and have other people be able to comment on that comment.

Ranked pvp is getting really ridiculously stupid at this point. 5 man premades against 5 man pugs is the biggest bullkitten I have ever seen since the new system was implemented. You could at least put a forfeit option if you are going to allow such unfair matches to occur.
It is bad enough half the time you get people on your ranked team that still don’t know how to play the kitten game.

Didn’t read the whole Thread, but i like the system as it is now. I think if you want to have a “solo only” queue it should be a real solo game mode. maybe with a few tweaks like “ashran” or what the name was in LoL. In all other cases u always depend on your teammates – an you never can really say with 100% certainty what your contribution to the win was. I.e. even if you killed the mos players- doesnt necessarily mean you saved the node for you team with these kills.
sorry for grammar, typos or anything- im German.

Justin ODell.9517:

Gabriell.4856:

Anyway to allow us to know our chance of winning before the match starts?

I would like to quit while I’m behind and get it over with as quickly as possible.

For ranked arena during a ladder, we plan to make the information available. I wouldn’t recommend leaving in one of those matches though.

That is good idea, but don’t forget to add description what it means. For example your victory odds is 15% you need to gain 300points to achive +1pt. Or even show the whole table with odds victory. Otherwise for many people it will not be clear what that odds means.

mysticsicness.7598:

Ranked pvp is getting really ridiculously stupid at this point. 5 man premades against 5 man pugs is the biggest bullkitten I have ever seen since the new system was implemented. You could at least put a forfeit option if you are going to allow such unfair matches to occur.
It is bad enough half the time you get people on your ranked team that still don’t know how to play the kitten game.

Those premades have inflated mmr meaning you’re playing against less skilled people. They have coordination, you should be simply better. Unless you are in the upper tiers of mmr you will face full premades vs all solos rather rarely. Plus, the work on that is still in progress.

Justin ODell.9517:

We also allow it because a lot of the time it doesn’t matter, especially now that we inflate party MMR by the number of members to compensate for the advantage.

Math: party.rating += party.rating * config.party.inflation * pow(party.size – 1, config.party.curve)

5 players with a mean of 1500 becomes 1845 with inflation at 5% and the curve set to 1.1.

There is still that occasional match though that gets totally ruined by the fail-safes. I’m not certain waiting longer for those players would improve quality for them or just reduce the number of games they get to play.

Edit: Correcting algorithm to use party.size – 1, which matches the LIVE code.
Edit 2: Fixed another issue in the way the algorithm was presented. Thanks, rotten.9753.

Twice vs same premade right now, both times i got full physical damage( full solo) against eles guard warrior.

How am i supposed to win with rangers, zerker warriors and thieves if it takes 2 of us to even kill one of them ?? Not that the result of the match would have been different, but still we can’t do anything…

kdaddy.5431:

For everyone complaining about solo please read all the posts

simply put there is a system based on your win percentage against the teams you plays rank. Im sorry but im willing to say the majority of you asking for solo que arent that good. Im also willing to say that your principles are much different then my own. Im sorry if 4 other people qued up and there good and im put in that group and we have to fight the abjured, ok bring it on. Let me have this chance, you literally have people on here complaining about fighting any pre-mades. Seriously what? You have other people trolling justin, really?

PVP is competitive, in what sport is the game completely fair? Seriously the same 4 guys win tennis every tourney, basketball has been dominated by 5 teams in the last 15 years. There is thousands of smucks playing this game and for some reason you come on here asking for a hand out when you can see they have a great system in place and is trying to get better.

Last note, no i will not wait any longer for ques. Ive lately had alot of 5+ min ques whether i solo or partially team que. If it goes back to the old days when we had solo and team ques ill do something else and at this point it might be wait for the expansion since theres nothing left for me to really do.

for great justice!

most people can’t accept they’re bad

nub: It’s the premades I tell you! I used to win all my games in solo q!
Justin: Only 7% of your last 500 games were against full 5 man premades
nub: that should never happen! your system is broken!
^^^^^^^^^
:)

So what are the odds that I queue solo (which is what I almost always do) and I get placed against another team entirely composed of solos?

Israel.7056:

So what are the odds that I queue solo (which is what I almost always do) and I get placed against another team entirely composed of solos?

depends on your MMR i would think
i would wager once you reach a certain MMR youre much more likely to be matched against premades then solos

oneply.9586:

depends on your MMR i would think
i would wager once you reach a certain MMR youre much more likely to be matched against premades then solos

So do we know what that MMR cutoff point is and how much is “much more?”

Also what about partial premades? Am I statistically more likely to face a 2 man premade than a 3 man premade?

Israel.7056:

oneply.9586:

depends on your MMR i would think
i would wager once you reach a certain MMR youre much more likely to be matched against premades then solos

So do we know what that MMR cutoff point is and how much is “much more?”

Also what about partial premades? Am I statistically more likely to face a 2 man premade than a 3 man premade?

paging Justin