Guild Wars 2

Ugh. The MMR system is an Elo system – specifically, it’s Glicko2, which is a version based on Elo that can adjust based on the size of the win, which Elo doesn’t.

Here’s the problem with an MMR-based leaderboard: you’re trying to put the top players up on the leaderboard, but you need those players to be keeping their hand in. Ranking systems like Glicko rank active players, and are only meaningful when compared against the pool of active players. A ranking of 1800, say, doesn’t mean much on its own. If the people at the top of the leaderboard stop playing, everyone else has to win much harder to get the same score as the current top players. That’s fine when matchmaking because you’re comparing active players to each other, but not for a leaderboard. The decay system was an attempt to fix this, but it’s fundamentally flawed because active and historical players can’t be meaningfully compared.

Going off W/L is a bad idea as well, because it’s trivially gamed: trash your MMR, or start a new account, before the season starts. You’ll be matched up against bad players over and over again, until the MMR system works out you should be higher.

What you want to show is who is the best player, as in the player that plays the best right now. Showing MMR ranking doesn’t work, because active and historical players can’t be meaningfully compared, but what we can do is adjust how many points you get for a win based on the team’s average MMR. This adjustment can’t be a linear function, to ensure that a player with an 80% W/L ranking down in the bush leagues can’t reach the top of the leaderboards over a ToL player with a 50% W/L rating who’s being matched against other top players.

50% win rate needs a nerf.

I think someone with a 50% win rate playing top players consistently deserves a spot on the leaderboard over someone with an 80% win rate crushing people who have never PvP’d before. The quality of the people you are playing and how well you perform against them should matter more than your win rate or how many games you are playing.

Malediktus.3740:

Make it so only your first 10 games of a day count towards the ladderboard.

Thats such a bad idea.
People who are abusing game mechanics will just play 10 games and loose every other game after this by beeing afk.
So they drop in mmr hard and have some easy matches the next day.
Have fun abusing this!

@topic
Nice to know, that i’m missing nothing by not playing
Keep up the good work/system!

@Josh
Why you need to test a system, that literaly everyone can tell you, is broken in its core?
So many people told you to go back to the pre december system and fix this. There were even good suggestions how to do this….
What do we have to do to be heared finaly?

I have the feeling that you are just testing, how long people will play Grind Wars 2.
The answer is: There will allways be people that will play it. PvE players love grinding, just look at WoW and all the asia grinders.
So i assume the system is working as intended?!

Regardless of LB, you still need to split queues into soloq and fullteam queue to truly represent “skill”.

butch.8136:

Regardless of LB, you still need to split queues into soloq and fullteam queue to truly represent “skill”.

There wasn’t any “skill” in the soloq leaderboard (being a person who hit rank 1 in it a few times (900 first time 1k ish 2nd …. forgot others). I might be a minority in terms of taking a break from the game, but the only thing I loved about soloq was that there was no inherent advantage in the other team (in our system now: duos/trios) it was a pure solo vs solo. You can make people believe they are playing solo by leaving party right after you accept match in the current system if you want to tryhard in a sense.

If anyone played many soloq matches they will tell you that: 80% of matches if you have your rotations down and one other person you are likely to win regardless of the 60% of your team that is bad higher up in ranking (depending on activity level). The only time you would get really good games is when activity is high and you face high rank players and the only people that determine the match are the 1-2 players on each team which made close games more than what I have experienced in the new matchmaking system. I met many good players playing high ranked soloq who I would ask if they wanted to team (or random people I met in solo asked me to team), but the new system now makes it so farm>skill and there is no sense of community to recognize good players, Anet shot their community in the heart with this “merged” matchmaking which only leaves the people that met each other in soloq in the previous system and a segregated “If I don’t know you, you suck. community”

An aside:

The system had flaws, but you were punished (in the old lb system) harder for a loss you were meant to lose. They (justin and co.) fixed the matchmaking system to match people better somewhat and removed the #1 reason why people hated soloq which was 4v5s and decided to replace it with unranked.

So far I’m liking this new MMR system. Please don’t change this. I think you are paired with teammates with a high W-L ratio. Before, you were “punished” for doing good. Before you will be paired with people with a lower win percentage. I don’t know what was the intention of arenanet with that, but the ones who are benefiting from it are not newbies that need nursing, it’s the point farmers who are abusing the old MMR.

Jasher.6580:

Flyphish.6398:

What is a loose?

http://www.youtube.com/watch?v=wFWDGTVYqE8

No, he asked what is a loose, not what is afoot…

Could use league of legends system of creating leagues which you are put into when you hit certain MMRs. That way it prevents people with low MMRs farming easy wins

Look at all these sick win/loss ratios. Working as intended?

Josh Davis.6015:

An important thing to keep in mind is that these events are tests, and the only thing we’re testing right now is the formula behind ranking. We’re gathering a lot of useful data that we can use to help us make better decisions. These “test seasons” not a final product by stretch of the imagination.

I know that test seasons are required to perfect your algorithms, but running a test season for 2 months and then having a long break before repeating the process. Come on, do you expect your PvP crowd to accept playing under such poor circumstances for half a year or more. While it’s a test to you and parameters you tweak for the optimal configuration, it comes across as broken to the actual players.

Why are you not running test seasons practically back to back, and shorten their duration to max 3-4 weeks. There are tons of suggestions to improve your algorithms – which you are quiet about and assumed to be making your own from the ground up rather than looking to well established ones. A test season needs not last longer than 3-4 weeks; especially at this early stage when you see the fallacy of games played so evidently already after just 2-3 days. That, hypothetically, some no-lifer can play and lose half of his 1000+ matches and top the table through sheer quantity gives the LB – and consequently you as PvP/eSports facilitator – little credibility.

A duration of 2 months is far too long – and as I assume you are implementing changes to your algorithms in parallel as you see the effects of the ongoing test season; there’s hardly a need for 2 months break between 2 test seasons. Your hardcore PvP crowd plays PvP every day, we can handle test seasons practically going back to back if you make tweaks and start a new test season after only 1 week break.

Add to that the Daily PvP objective of ‘Daily <class> winner’ which sees a great imbalance – and even isn’t distributed equally among the classes throughout the month. Rework it to simply winning a ‘Daily PvP win’ (heck, make it Win 2 matches) and you’ll have players spreading out across the profession scale.

And where is Dishonor – how can it be that I have to suffer through 4v5 games where 1 person has clicked ‘Ready for match’ then just go AFK for half a match once the map has loaded. Where’s the auto-kick and replacement, where’s the penalty to be given to the player who creates a bad PvP experience for 4 (or even 9) other players?

Another contributing factor is clear class imbalances that persist still after many, many weeks/months of being reported. A Ranger only needs two keys these days (Longbow #1 and #2), or take the Turret Engis, or even the base damage of the Lich Form for Necros.

Also there’s the whole case of the actual MMR system and the impacts of wins vs games played on the Leaderboard, which others have discussed profusely above and in numerous other threads.

I agree 2 months are too much, but the alternative now would be off-season after 1 month (where lb is really useless).

I really think the next season will be with HoT (q3). I hope in a league season, separated queue for stronghold and each game mode and not super grind leaderboard (pretty curious how guild lb fits in the system).

I’m just wondering why they’re not testing a system that actually WORKS instead of one that has been proven time and time again to not work.

Tranassa.4968:

Malediktus.3740:

Make it so only your first 10 games of a day count towards the ladderboard.

Thats such a bad idea.
People who are abusing game mechanics will just play 10 games and loose every other game after this by beeing afk.
So they drop in mmr hard and have some easy matches the next day.
Have fun abusing this!

Easy to fix, too. Freeze MMR after you played your 10 games.

Maybe I’m missing something, but why do you need to “test” something that should be obvious. How do you not see that this algorithm is primarily going to just be a list of top grinders???

You originally had a good algorithm, but just had to fix the obvious problem with decay.

Why are you turning a simple solution into rocket science?

SPESHAL.9106:

Maybe I’m missing something, but why do you need to “test” something that should be obvious. How do you not see that this algorithm is primarily going to just be a list of top grinders???

You originally had a good algorithm, but just had to fix the obvious problem with decay.

Why are you turning a simple solution into rocket science?

Grinders you say. Not only that, but just to make sure the grindres stay on top, this is the type of teams your facing in rank. After i swore i would’nt play ranked & i did, i obviously played vs multiple stacked meta teams. (3 meta warriors -2meta Eng) Or (3 Eng- 2 Warriors) & so on……………………… ArenaNet, Waa waa waaaaaa! :s

Ps: worst part, i’m sure they wher not even premade, just players playing all the Same top Metas. No way they gonna play a mesmer on Rank.. God Forbid!

MarkPhilips.5169:

I agree 2 months are too much, but the alternative now would be off-season after 1 month (where lb is really useless).

I really think the next season will be with HoT (q3). I hope in a league season, separated queue for stronghold and each game mode and not super grind leaderboard (pretty curious how guild lb fits in the system).

Off-season can easily be countered either by

1) While there is need for testing, new test season starts 1 week later. Yes, it puts pressure on the devs, but kitten it, they should be spending the tournament’s duration to make changes and then they have a short time to introduce these before a new round of testing. Prolonged testing with your live end user audience with months of off-season while nothing happens anyway to what’s been proven broken during the tournament simply leads to crap product and negative community not believing in the product.

2) Introduce automated tournaments; let tournaments run back to back. Make changes as you go so these are queued for the next tournament, and if you don’t have time to put in all you wanted, then put in smaller tweaks to see the effects of these and schedule bigger changes for the next season. That way you at least get continuous tournaments for the PvP player base; and because there is no down-time on the competitive seasons, you also avoid situations like the people now having worked hard to farm their way to top of the leaderboards in what was an offseason break between mid-January and mid-March.

Chaith.8256:

• This leads to the realization that medium to high MMR players can only attempt to achieve quality matches, with a competitive win rate representative of their skill, without severe handicaps, when they queue as a full group. This further limits group playtime to a couple hours a day, where a group of 5 competent players can be organized and continuously play. This creates a further disadvantage in win farming compared to low MMR solo queuers.

The real true.
In my case I’m only playing with full group in ranked during the prime time. I’m getting really cool matches. But its impossible to achieve a good rank, I’ll stay in % forever with this algorithm.

Vissarion.6509:

EU is even worse:

1st: wins:68 looses:46 win%:59.65%
2nd: wins:63 looses:16 win%:79.75%

And how do you know this is worse?

As long as solo queues and full try-hard TS and Meta coordinated premades are not seperated player 1 has probably been solo-queuing with random non-optimized teams of 3 rangers, 1 thief and a necro and thus having a low win chance compared to player two who played with his full meta-proof team that is fully cooperating on TS each match.

Why would it be fair that a fully coordinating team with high MMR and a full group would gain the same kind of points for a win as a team of random solo-queuers would get if they would miraculously manage to win against them?

I pretty much always solo-queue and you’d be surprised how many time I meet premades while being in a team of ranger-ranger-thief-mesmer/necro + me who don’t even say “hi” when you greet the team at the start, let alone try to coordinate.

Malediktus.3740:

Tranassa.4968:

Malediktus.3740:

Make it so only your first 10 games of a day count towards the ladderboard.

Thats such a bad idea.
People who are abusing game mechanics will just play 10 games and loose every other game after this by beeing afk.
So they drop in mmr hard and have some easy matches the next day.
Have fun abusing this!

Easy to fix, too. Freeze MMR after you played your 10 games.

Still not good, cause you basically have to play every day to have a chance to be on top. Just one day without playing and you can’t come back to the top.

And freezing the MMR means ppl can troll, and if they can they will do.
You should know that. Or you forgot about all the trebuchets “Mr. X” built to grind some AP?

The current Grindboard is the problem!

Great post, couldn’t have written this better myself

Wait, are you telling me that Anet has been working on gw2 esports for 2.5+ years, but they’re leaderboard system is based on grind and not mmr?

I don’t know whether to be sad, angry, or amused. I really hope everyone that continues to play pvp does so because they enjoy it, and has 0 expectations of it getting any better. I’d say this is a wakeup call but let’s be honest, this is typical Anet.

Hopefully anet does implement a skill based leaderboard, however i reckon they would also have to “unmerge” solo queue and team queue.

This blows my mind. It has been a couple of months since the last test. You are telling me that since that time, they couldn’t figure out a better way to do the leaderboard? Someone at ANET actually thought this new algorithm was a good idea?

Wow, just wow.

Flumek.9043:

• M

• M

• R

^This.
As long as there is a Ladder/point per win/loss based leaderbord it will always, no matter what you change and do be heavily dependent of the number of games you play. Even more so if the math thingy behind all this is aimed to get people to a 50% w/l ratio for better matches while at the same time you seem to gain more points per win than you lose per loss. The only purpose of such algorithm is an overly complicated way to count matches played (with some errors even! You could do better and easier btw – ask any 1st grade school kid about how addition works – they might be able to help out). Of course I understand that it is encessary to put some impact on matches palyed or we’d see 1win:0loss people dominate the rankings – but this can’t be it. If we wanted a ranking for time devoted to this match we’d look at the AP leaderboards (or that weird chinese VIP emporer title). I guess most of us regular pvp players (be it high mmr or lower like myself) who enjoy a competetive gamemode agree that the old leaderbords were a lot better than this.
Why are we not getting it back then? I honnestly believe the main issue is simply that noone at Anet seems to understand what PvP stands for. PvP stands for palyer versus palyer for a reason: It is about competition – and about comparing each others skill at the game. And this is definitely not meassured through the time played but simply by winning vs others. And the better those enemies are the higher you should be on the leaderbords. The big surprise however is, you actually already do have the perfect sorting system: MMR! Add some decay to it and you got a perfect viable leaderbord.
From what I have read in some other thread, some people at Anet seem to be afraid of toxity and elitist behaviour if MMR was made public in any way. But hey, those are the people who did not get the sense of PvP at all. There already is elitist behaviour and toxity anyways, it won’t get worse just ebcause all of a sudden people see the truth. It will deillusionize some, make some people very happy and end the discussion where people shout at each other about who was better. As already mentioned, it is about competing and comparing with other Players. Understand that or you might as well replace PvP with some sort of speed clearing dungeons or anything, because what we currently have is just half the PvP.
Right now we only have real competition and sort of rankings for highest tier palyers through the weekly cups. I even got a tiny little hint for you in case you are scared to scare off the mid and low tier players by having MMR based leaderboards: Show more than 1000 names on it. Then people can comapre to their friends and other palyers they get matched against regularely and will be able to actually see themselves improve. I’m fairly sure they’d enjoy that more than pointless grinding (I would! No matter if I was working to get to rank #1 or #1500 as next goal). As for rewards: Add soemthing like for the WvW seasons to the PvP ones. Some sort of cheap titles the grinders and PvE’ers can brag with and some useless reward box with a couple of random rares, greens, whatever in it and then add good rewards for the leaderbrod’s top people. Hell, you could even award people for climbing a certain amount in the eladerboards of their previous position during a season. Once the PvP population increases a lot, add divisions and everyone will be able to have access to the “winners” rewards in their division. That, along with the regular reward paths should be enough to give people motivation to play PvP. And on a side note, do not underestimate the influence of a good leaderbord on peoples motivation.

Problem with going for a win/loss ratio is also that mathematically, it’s not really possible to keep it up in solo queue. Say you’re a really good player. That means you have 1/5 chances of being a strong team. If the 4 other guys are horrible, no matter how good you are, you’ll lose. So that’s you here, depending on how lucky you get, gearing towards an average between 50 and 70%, leaning more and more towards 50% cause the more you’ll play, the more the odds are gonna even themselves. It’s like a coin toss. Toss it ten times, you might get 8 heads, 2 tails, toss it 100 times, you’re more likely to get around 50 heads, 50 tails. Of course, premades can raise that potential, but then again, if they’re gonna benefit from grouping, maybe there should just be a premades arena and a solo arena cause those guys get an unfair advantage from grouping. So all in all, if you solo queue, 59% isn’t bad at all. Statistically, unless you were born with a shamrock in your kitten, you can’t do much better. More points for beating a high ranked group, that’d be the beginning of something.

I’m pretty sure they didn’t even reset MMR for this new leaderboard test.

What a complete joke.

Reikou.7068:

I’m pretty sure they didn’t even reset MMR for this new leaderboard test.

What a complete joke.

The devs see resetting MMR as a disruptive and bad experience for the PvP population because people would be complaining about getting bad matchups until the MMR was re-issued properly.

What they don’t take into account is the huge imbalance that MMR has on leaderboard point generation.

Chaith.8256:

The devs see resetting MMR as a disruptive and bad experience for the PvP population because people would be complaining about getting bad matchups until the MMR was re-issued properly.

What they don’t take into account is the huge imbalance that MMR has on leaderboard point generation.

I guess at the end of the day this is one of the big reasons why we’re not seeing an MMR-based leaderboard…

Still, what a complete joke.

feel like amulets need to be taken into account as well as class. just had a game with four zerkers who all insta downed to focus, with me (valk shout warr) the only person who could stand on point. i knew instantly from the team comps we had lost, and we did. zerker should be limited to three per team. four zerker fails without TS.

Reikou.7068:

Chaith.8256:

The devs see resetting MMR as a disruptive and bad experience for the PvP population because people would be complaining about getting bad matchups until the MMR was re-issued properly.

What they don’t take into account is the huge imbalance that MMR has on leaderboard point generation.

I guess at the end of the day this is one of the big reasons why we’re not seeing an MMR-based leaderboard…

Still, what a complete joke.

Actually, MMR greatly affecting point farming rate I think it’s a big reason why points based leaderboards aren’t functional in any way, and MMR based leaderboards (or anything but the darn points based style) is better.

Arenanet has a design goal for the leaderboards. That is so that a player who increases games played will progress the leaderboards faster. This design goal must be abandoned. It’s the anti-thesis of what a leaderboard should be. Players should never be limited in the games they can play during a season, but if the focus is on grinding more than dominance, as it is by a huge margin, right now, leaderboards are not going to be engaging for anyone except for the 2 people poopsockin’ per region.

The original solo queue leaderboard was flawed but much, much better than the current iteration. At least I could track my relative progress and games did start to get more reliable once in the top 500 or so.

Now I have nothing to track my “progress”. It is not feasible for me to play 30+ matches a day. Even getting in the top 1000 to use a third-party site is a bit of a grind.

I like seeing daily W/L. I like knowing whether I am improving or not. I like seeing familiar faces on my team. I had all those things with the old leaderboard.

PvPers in all games like having a meaningful gauge of relative standing. That is non-existent now. I am not trying to be in the top tier, I just want relative progression. That is gone now. Further, the new combined queue and grind-based leaderboard incentivizes bad behavior.

Everyone is now checking the party status of the opposition, complaining about their teammates, and just in general people are not enjoying the game as much as before. Combine all these factors with daily profession achievements and you have the perfect recipe for a disgruntled, disengaged community.

The first step is fixing the leaderboard, so players of all abilities still have something to track.

The second step is separating queues so people can again take accountability for their play and not blame it on facing a premade. Concurrently, solo players will be guaranteed to face other solos, teams can take pride in beating other teams.

The third step is to remove profession dailies. This leads to strange team compositions and people playing professions they should practice with before jumping in ranked. The system as it stands incentivizes people just playing ranked with alts to get the dailies. How else is one to keep up with the new leaderboard?

Flyphish.6398:

The original solo queue leaderboard was flawed but much, much better than the current iteration. At least I could track my relative progress and games did start to get more reliable once in the top 500 or so.

Now I have nothing to track my “progress”. It is not feasible for me to play 30+ matches a day. Even getting in the top 1000 to use a third-party site is a bit of a grind.

I like seeing daily W/L. I like knowing whether I am improving or not. I like seeing familiar faces on my team. I had all those things with the old leaderboard.

PvPers in all games like having a meaningful gauge of relative standing. That is non-existent now. I am not trying to be in the top tier, I just want relative progression. That is gone now. Further, the new combined queue and grind-based leaderboard incentivizes bad behavior.

Everyone is now checking the party status of the opposition, complaining about their teammates, and just in general people are not enjoying the game as much as before. Combine all these factors with daily profession achievements and you have the perfect recipe for a disgruntled, disengaged community.

The first step is fixing the leaderboard, so players of all abilities still have something to track.

The second step is separating queues so people can again take accountability for their play and not blame it on facing a premade. Concurrently, solo players will be guaranteed to face other solos, teams can take pride in beating other teams.

The third step is to remove profession dailies. This leads to strange team compositions and people playing professions they should practice with before jumping in ranked. The system as it stands incentivizes people just playing ranked with alts to get the dailies. How else is one to keep up with the new leaderboard?

Engaging the community.

That’s a massive take-away.

The developers were trying for exactly that when they introduced the ‘play more to advance faster’ approach. They felt that it would stimulate everyone to increase the volume of games played.

However, it only succeeded at completely disengaging all mid-high tier players from leaderboard progression, as well as every low tier player who can’t solo q into ranked for over 8 hours a day.

Low MMR players who solo queue ranked for 8+ hrs a day are the only ones who are engaged in mastering the current leaderboard progression system. Clearly the data is showing that.

When the top 3 guys on the leaderboards are literally rated 50-100% higher than the top 10-20 guys, and there is not a high MMR player in the top 100, you know there are some problems with who’s being rewarded.

I donno, I just hope they get rid of that terrible decay system too. I played very very casually and I was often teamed up with terrible team mates or against terrible teams, I had like 70 to 80% win ratio, I had to carry hard most of the time, that was totally unfun as well. There was no way I could ever reach high positions, of course.

I stopped trying and I just went to unranked with non-meta builds and with classes I don’t handle as well. I stopped playing entirely, eventually.

The system will work better if it was tpvp only, instead of solo players thrown into the mix…

Teams are easier to track. They act like 1 player. They will almost always perform the same with very little outlying results.

Solo players will have mixed MMR because they are all rated between each other. Your rating should be the average of the best players and worse players on your team. As this averages goes on, it will eventually tell you what your ratingn is depending on how high, or low, this average tend to lean. It’s only after 1000 games will you see a difference between one player and the other. This is why MMR is so screwed up. Players are not playing large amounts of games, (nor do we want to) and we’re often playing the with and against the same players. The time invested to make the system work is too low. Not to mention certain KEY elements are skewing the results. Like build swapping between meta and off meta, heavy casual play, bad matchmaking due to low population, etc.

My point is, maybe Anet are setting themselves up for failure because key features havent been in place, or quite simply, they’re using a system that doesnt mesh with this game very well.

So basically they’ve made no progress after 5 months time. It’s still all about farming wins. Just wow.

Well done Anet, I see you’re still going at Anet pace.

RockSteady.7123:

So basically they’ve made no progress after 5 months time. It’s still all about farming wins. Just wow.

Well done Anet, I see you’re still going at Anet pace.

It is worse than no progress, it has regressed.

Guys, I really can’t understand why you bother with complicated algorithms when you want to reward most the activity. With the same success you can just make the leaderboards with the total amount of the games played. Then it’ll be the same grind fest like now, and even a lil bit more informative about the skill of the players, right?

Merus.9475:

Showing MMR ranking doesn’t work

Using MMR as a ranking system doesn’t work well because it’s so volatile; players that open on a hot streak pop up to the top of the leaderboard, where they’re incentivized to camp out on that high rating.

The cleanest way to handle it for a system like Glicko2 is to make the leaderboards based on MMR – 2*RD; basically, instead of basing it on the point estimate (which can be really volatile), it’s based on the level the system is ~98% confident that you are better than.

Handling rankings in that way has a couple of desirable features. First, more games at the same skill level tends to move players up the board, as even though their MMR isn’t increasing, their RD is decreasing as the system gets more and more confident of their skill level. Second, players that open on hot streaks don’t shoot up the leaderboard, as their RD remains high until they’ve built up a meaningful sample of games.

Plus, RD decays pretty quickly with inactivity (at least, it should), which drops people down the boards if they aren’t currently active without hitting their MMR.

The team queue wins/losses is a different problem – I imagine it’s perfectly solvable with the data A.Net already has, but it’s not a trivial statistical problem. Premade effects are generally not homogeneous, and matchmaking rating in general has issues with heteroskedasticity that don’t necessarily fit a clean form.

what if an account is only added to the leaderboard once they pass a volatility threshold for their mmr. new accounts start with higher volatility to help them find their placement quicker, and the volatility is a function of games played, time since last game played, and win/loss of the last 50-100 games played. then also add permanent slowish decay so you have to keep playing to keep your position. and just use mmr as the leaderboard score.

Personally I don’t mind whether the leaderboard is MMR or point based. However, problem with this system is that it’s simply… unfair. Now you are actually punished for being a “good” player. When your MMR is high you will pretty much get only 1 point per win no matter what, while lower MMR players (who are actually way higher on the leaderboard!!!) will often get more points per win. That is just ridiculous.

Point based leaderboard should be same for everyone.
MMR based leaderboard is straight forward, like it was before.

Problem is that they are trying to combine both. It just doesn’t work. I think players should always be matched against each other based on the leaderboard position. That position is supposed to show good the players are, so then they could fight against players who have similar skill level. Using MMR for matchmaking only and not for leaderboard position makes no sense at all.

insanemaniac.2456:

what if an account is only added to the leaderboard once they pass a volatility threshold for their mmr. new accounts start with higher volatility to help them find their placement quicker, and the volatility is a function of games played, time since last game played, and win/loss of the last 50-100 games played. then also add permanent slowish decay so you have to keep playing to keep your position. and just use mmr as the leaderboard score.

That’s only useful for Team Queue. This will have a number of issues when tpvp and spvp is merged together.

well noone cares bout leaderboard…everyone is already full of worthless lamas anyway so there’s not even a tiny little reason to give any kitten…let alone it’s all about how much ppl play, even if they totally suk

noone outside pve tryhards is gonna setup a full rooster wasting time both waiting for people and qtimes for nothing…you just soloq trying to carry as much as possible while your team is 3v1ing on close, dieing at svanir, going for lord or something else…but hey that’s pvp playerbase right now…so you already know how is gonna end when you que up

luckily you can still go afk and browse youtube for more interesting stuff..looking at screaming goat’s videos is way better then looking at ppl 3v1ing a shout war on close while 4th guy is going chieftain…(Probably getting rekt ofc)

Here’s my experience with the current ladder:

I’d love to earn a Llama, I just don’t have the time to put 10+ games of ranked PvP in a day. Therefore, I can’t get a llama because I can’t place on the LB. My biggest concern, though, is really the matchmaking algorithm in use right now. I think the thing that’d help a player like me the most is the ability to find a team or a group of players I can play with when I queue, because matchmaking usually pairs me with lower MMR players to counterbalance my mid-high MMR, and as a result I have matches that really aren’t fun where I have to carry hard or get smashed by mediocre players slaying 4 players from my team with 2 players from their team.

Honestly i think the issue with the actual LB system is much worst than Match up issues.

Josh Davis.6015:

Thanks for writing up your concerns so succinctly.

This test season is intended to test an algorithm that’s tailored towards ranking players higher for better W/L ratios, in addition to using points gained from playing (meaning games played does not become valueless). The algorithm we’re using this time around is completely different than what was previously tested, and they both have different purposes. We’re already having internal discussions on how we can adjust this season’s algorithm to better suit the goals we stated on Thursday – rewarding players who play well, not just play. These discussions will continue into the next week, and now we have an entire weekend’s worth of data that we can look at and extrapolate on.

An important thing to keep in mind is that these events are tests, and the only thing we’re testing right now is the formula behind ranking. We’re gathering a lot of useful data that we can use to help us make better decisions. These “test seasons” not a final product by stretch of the imagination.

is gw2 still in beta? after 3 year we are still stuck into test… and still the matchmaking not working like class stacking , kitten meta and a leaderboard where korean>skill
gz like always

I think the fundamentals of everything are quite fine except there should be a maximum amount of games that count toward leaderboard ranking. People are fixated too much on MMR. I agree it should be shown but there are many factors outside of a players control that can effect your MMR especially if you solo. For one thing we don’t have leagues and in low population times you can get matches against groups and partial groups quite often. That is what win probability attempts to address. If you are winning the matches you should be at a high clip what more can you do? They just need to cap the games at something reasonable that rewards playing the game but not obsessively. I think 30-40 games a week is a sweet spot for maximum that count that could be debated.

People taking shots at solo queue people with 60% win rates are out of there mind. Just because you run in a group all the time and win 70-80% of your games that imo is not more difficult then soloing all the time and pulling off a 60%+ win ratio. I lean toward the 60% solo being more difficult. A full team can and does carry a lot of people. We know how a lot of teams work is there are a few solid players and then a couple of stars that make or break it.

Saiyan.1704:

insanemaniac.2456:

what if an account is only added to the leaderboard once they pass a volatility threshold for their mmr. new accounts start with higher volatility to help them find their placement quicker, and the volatility is a function of games played, time since last game played, and win/loss of the last 50-100 games played. then also add permanent slowish decay so you have to keep playing to keep your position. and just use mmr as the leaderboard score.

That’s only useful for Team Queue. This will have a number of issues when tpvp and spvp is merged together.

why? the old team queue is the same as the current ranked.

While everyone is throwing ideas out there, here’s what I think, ranked should only be for 5 man premade team, when you go ranked you know want a fair 5v5 both team with voice comm etc. You also know your performance will be ranked, that is were everyone will tryhard and no one will complain if they are facing a premade. Having that said, to make ranked 5v5 only you need somewere for other players to play. Which is why I think they should remove the kittenty maps that no one wants to play from unranked (courtyard, skyhammer and spirit watch). Then if the ranked is 5v5 only, you can remove the mmr like Reikou said in a earlier post.

Rising Dusk.2408:

Here’s my experience with the current ladder:

…..My biggest concern, though, is really the matchmaking algorithm in use right now. I think the thing that’d help a player like me the most is the ability to find a team or a group of players I can play with when I queue, because matchmaking usually pairs me with lower MMR players to counterbalance my mid-high MMR, and as a result I have matches that really aren’t fun where I have to carry hard or get smashed by mediocre players slaying 4 players from my team with 2 players from their team.

honestly i dont think the matchmaking is bad. I get alot of close/even matches. I dont mind playing with players worse then me as long as the enemy team has them aswell. Obviously sometimes you get bad matches, but i think thats more due to the small player base. The matchmaking algorithm seems fine to me. The Ladder is the problem. We need a Ladder showing MMR. This will btw also increase the overall matchmaking too. It leads to more teams forming and playing —> less premade vs soloq