Guild Wars 2

Chaith.8256:

Novuake.2691:

It wouldn’t be easy to get it right, but how would you do it otherwise? How else will would you promote individual players out of lower tiers in the short period a season lasts?

So the goal is to give additional pips to individual players based on merit, but I’m not convinced that is necessary to having a skill based league system. It might even create problems, like division getting way too high for their skill, via boosted rewards for beating on minionmancers & eles all game and accomplishing nothing.

I have a rather simple proposal to change the matchmaker so this problem doesn’t happen in the first place, and that’s simply ship a priority change of how MMR grouping works.

In any pip range we have a variety of MMRs, let’s group them into 3 camps – below average, average, and above average.

Right now, from my understanding of the matchmaker, we’ll have a team of 5 taken from the below average camp and randomly faced off against a team from the below average, average, or above average camp.

When ArenaNet heard our desires of having teammates relatively close to us in skill, this MMR grouping was absolutely NOT what we wanted, if I can be so bold as to speak for the general population.

The matchmaking in season1 was good in the way that it had high quality, close matches. A big problem (excluding other problems) is that it did whatever it took to even the team MMRs, including adding players who didn’t belong there, to handicap, or boost the team MMR. This was disruptive.

What I propose is a combination of S1 and S2 matchmaking, where matchmaking still determines your matchups based on pip range, like it does now, to stay in line of ‘perserving prestige’ in divisions.

The main difference would be instead of finding two very different groups of 5 alike MMR people, and having a potentially imbalanced match, try to find the 10 most similarly rated players in MMR within the pip range, rank them #1-10 based on MMR, and have even vs. odds. With an expanding PIP range, and variation in MMRs accepted as the queue goes on.

In a 5v5 with the 10 most similarly rated players, there won’t be as big a need to promote certain players over others, as it’ll be much harder for you to become underrated in MMR by the system.

I could even TLDR; and say bring back the matchmaker from the era of old Leaderboards & Solo queue days. At both extremes of skill, the best players had 70% winrate, and the worst players had about a 45% winrate. Right now I’d ballpark the two extremes at a respective 85% and 20% winrate.

With the league infrastructure we have now, and a few more tweaks, we can have a pretty awesome casual PvP league. The problem was that after we invented the wheel, we invented and used a square wheel with points/grind based leaderboards, and then a triangular wheel – In S1, getting high divisions without fighting tough enemies – and now we’re using a rectangular wheel in S2 – grouped up based on MMR and face different MMR groups!

I’m a bit weary of all the crazy matchmaker/ladder fads we’ve tested. I’m happy we added the league system which was a huge step forward, but we just need to get a normal and ordinary matchmaker to complement it like we had before.

+1 …

It’s a good reply Chaith but I really think in a game like GW2, where professions are really different and balance is really important, A MMR – league system is not enough to guarantee a good game.

Profession/builds are really important, there should be some control system on balancing the right mix sustain/damage in both of teams, or if you have a bad mix of profs against very good comp is always an handicapped game.

Phil.8901:

It’s a good reply Chaith but I really think in a game like GW2, where professions are really different and balance is really important, A MMR – league system is not enough to guarantee a good game.

Profession/builds are really important, there should be some control system on balancing the right mix sustain/damage in both of teams, or if you have a bad mix of profs against very good comp is always an handicapped game.

Agree to disagree,

Players choosing to not run elite specs, players choosing to not switch characters so there’s a semblance of balance of support and damage on a team, that is on the players to decide. Impossible for the system to manage, outside of simply assigning the poorly adjusted and unwilling to switch players a lesser rating.

In high MMR games I see players playing the right things and switching their picks. Your ability to play strong and well adapted builds, multi-class & pick a comp properly would simply be something that is characteristic of higher divisions if it were to be implemented properly.

Not always people want to switch profs or they’re good at the same level with every profs.

A good pvp game needs to guarantee good quality games at every level, not only in diamond/legendary.

Locking builds and roles should be prioritized like design, the problem is in this game many times ANET (balance team) doesen’t really know which role they want to give to some prof.

There is no clear direction about balance, this is one the main issue of this game.

Phil.8901:

Not always people want to switch profs or they’re good at the same level with every profs.

A good pvp game needs to guarantee good quality games at every level, not only in diamond/legendary.

Locking builds and roles should be prioritized like design, the problem is in this game many times ANET (balance team) doesen’t really know which role they want to give to some prof.

There is no clear direction about balance, this is one the main issue of this game.

I’ve seen lots of PvP games that lock the class picks, but in GW2 we need the ability to switch characters really bad, for more than one reason. When I play SMITE I’ll see someone pick a tanky bruiser god and watch him play/build it like an an assassin to the detriment of the team. It’s never foolproof. Fortunately the players who can’t make decisions to save their pips will have less pips than everyone else.

Also it’s funny you mention that the lines between roles seem blurred for professions in PvP, I think they couldn’t be more cut and dry.

I hardly ever see a profession using a build that I can’t use the same two words to describe it as I do with the rest of the profession.

Every Ele: Support bunker.

Every Engi: Support bruiser.

Every Ranger: Support & Assassin-Pet

You get the idea.

If you mean ArenaNet can’t decide whether to make casual Thief/Warrior players a little bit silly, or an outright free kill, then I suppose you’re right. This just reinforces the need to have a backup profession or two, or you and your team might pay the price in pips.

But I seriously think that balance and pre-game picks are not so huge an issue that we can’t talk about having nice matchmaking.

Chaith.8256:

Novuake.2691:

It wouldn’t be easy to get it right, but how would you do it otherwise? How else will would you promote individual players out of lower tiers in the short period a season lasts?

So the goal is to give additional pips to individual players based on merit, but I’m not convinced that is necessary to having a skill based league system. It might even create problems, like division getting way too high for their skill, via boosted rewards for beating on minionmancers & eles all game and accomplishing nothing.

I have a rather simple proposal to change the matchmaker so this problem doesn’t happen in the first place, and that’s simply ship a priority change of how MMR grouping works.

In any pip range we have a variety of MMRs, let’s group them into 3 camps – below average, average, and above average.

Right now, from my understanding of the matchmaker, we’ll have a team of 5 taken from the below average camp and randomly faced off against a team from the below average, average, or above average camp.

When ArenaNet heard our desires of having teammates relatively close to us in skill, this MMR grouping was absolutely NOT what we wanted, if I can be so bold as to speak for the general population.

The matchmaking in season1 was good in the way that it had high quality, close matches. A big problem (excluding other problems) is that it did whatever it took to even the team MMRs, including adding players who didn’t belong there, to handicap, or boost the team MMR. This was disruptive.

What I propose is a combination of S1 and S2 matchmaking, where matchmaking still determines your matchups based on pip range, like it does now, to stay in line of ‘perserving prestige’ in divisions.

The main difference would be instead of finding two very different groups of 5 alike MMR people, and having a potentially imbalanced match, try to find the 10 most similarly rated players in MMR within the pip range, rank them #1-10 based on MMR, and have even vs. odds. With an expanding PIP range, and variation in MMRs accepted as the queue goes on.

In a 5v5 with the 10 most similarly rated players, there won’t be as big a need to promote certain players over others, as it’ll be much harder for you to become underrated in MMR by the system.

I could even TLDR; and say bring back the matchmaker from the era of old Leaderboards & Solo queue days. At both extremes of skill, the best players had 70% winrate, and the worst players had about a 45% winrate. Right now I’d ballpark the two extremes at a respective 85% and 20% winrate.

With the league infrastructure we have now, and a few more tweaks, we can have a pretty awesome casual PvP league. The problem was that after we invented the wheel, we invented and used a square wheel with points/grind based leaderboards, and then a triangular wheel – In S1, getting high divisions without fighting tough enemies – and now we’re using a rectangular wheel in S2 – grouped up based on MMR and face different MMR groups!

I’m a bit weary of all the crazy matchmaker/ladder fads we’ve tested. I’m happy we added the league system which was a huge step forward, but we just need to get a normal and ordinary matchmaker to complement it like we had before.

+1, You hit the nail on the head.

Chaith.8256:

Novuake.2691:

It wouldn’t be easy to get it right, but how would you do it otherwise? How else will would you promote individual players out of lower tiers in the short period a season lasts?

So the goal is to give additional pips to individual players based on merit, but I’m not convinced that is necessary to having a skill based league system. It might even create problems, like division getting way too high for their skill, via boosted rewards for beating on minionmancers & eles all game and accomplishing nothing.

I have a rather simple proposal to change the matchmaker so this problem doesn’t happen in the first place, and that’s simply ship a priority change of how MMR grouping works.

In any pip range we have a variety of MMRs, let’s group them into 3 camps – below average, average, and above average.

Right now, from my understanding of the matchmaker, we’ll have a team of 5 taken from the below average camp and randomly faced off against a team from the below average, average, or above average camp.

When ArenaNet heard our desires of having teammates relatively close to us in skill, this MMR grouping was absolutely NOT what we wanted, if I can be so bold as to speak for the general population.

The matchmaking in season1 was good in the way that it had high quality, close matches. A big problem (excluding other problems) is that it did whatever it took to even the team MMRs, including adding players who didn’t belong there, to handicap, or boost the team MMR. This was disruptive.

What I propose is a combination of S1 and S2 matchmaking, where matchmaking still determines your matchups based on pip range, like it does now, to stay in line of ‘perserving prestige’ in divisions.

The main difference would be instead of finding two very different groups of 5 alike MMR people, and having a potentially imbalanced match, try to find the 10 most similarly rated players in MMR within the pip range, rank them #1-10 based on MMR, and have even vs. odds. With an expanding PIP range, and variation in MMRs accepted as the queue goes on.

In a 5v5 with the 10 most similarly rated players, there won’t be as big a need to promote certain players over others, as it’ll be much harder for you to become underrated in MMR by the system.

I could even TLDR; and say bring back the matchmaker from the era of old Leaderboards & Solo queue days. At both extremes of skill, the best players had 70% winrate, and the worst players had about a 45% winrate. Right now I’d ballpark the two extremes at a respective 85% and 20% winrate.

With the league infrastructure we have now, and a few more tweaks, we can have a pretty awesome casual PvP league. The problem was that after we invented the wheel, we invented and used a square wheel with points/grind based leaderboards, and then a triangular wheel – In S1, getting high divisions without fighting tough enemies – and now we’re using a rectangular wheel in S2 – grouped up based on MMR and face different MMR groups!

I’m a bit weary of all the crazy matchmaker/ladder fads we’ve tested. I’m happy we added the league system which was a huge step forward, but we just need to get a normal and ordinary matchmaker to complement it like we had before.

I think your proposed system is how S1 worked — find players based on pip range and with similar MMRs, and then split them into teams such that each team has similar MMR. This means that below-average players could reach legendary simply by beating other below-average and average players. And above-average players would have to constantly beat other above-average players to reach legendary. The system really breaks down when there aren’t enough people playing at a given time, which causes the matchmaker to expand the MMR range such that good players wind up having to carry terrible players.

I personally would rather see a slightly modified (smarter) S2 system with the following changes:

• Increase RD (rating deviation) slightly for purposes of forming a team —* This gives “average” players a higher chance to team with “above average” players, or “below average” a higher chance to team with “average.” So having a high MMR is less a guarantee of victory, and a low MMR less a guarantee of loss.

• Tie MMR loss to Pip Loss —* The only regular complaint I see with S2 is about people claiming that they’ve fallen into “MMR hell”. This only happens if you’re constantly losing games (so losing MMR) without losing pips (so not dropping down and playing with worse players). This could easily be resolved by adjusting the system so that you only lose MMR when you lose a pip. In other words, if you are 0 pips into Ruby, further losses won’t decrease your MMR. Voila, no more MMR hell.

• Smarter MMR system —* The amount of MMR you gain/lose after a game should be affected by the final score. For example, the MMR impact for a 50-500 loss vs an equal-MMR team should be greater than the impact if you lost by only 495-500 to that same team. And the system could even increase your MMR if you had a close loss vs a much stronger team. This may incentivize players to try harder even when they think the loss is inevitable. (This ties in with the next suggestion)

• More transparent MMRs* — Let players see their personal MMR. Also, display the average MMRs of each team after a game. This way, players can track their own growth (or regression), and also better understand how their games were lined up. This would be particularly helpful for players who incorrectly feel like they’re stuck in MMR Hell. (For example, several of the MMR Hell gameplay vids that have been posted show that the poster was actually on the stronger team, but the poster simply lacked the awareness to realize it).

Nah, slightly modified in the wrong way?? Just give people MMR vs MMR or pip vs pip range. That’s what lol, data or Smite or CS:GO uses. So what aren’t we doing that?? Even the Glicko 2 excel add on won’t let me match a 1000 vs a 3000 chess player.

Start with the basic MMR vs MMR (within 2 tier) or pip vs pip( this one is risky though) and go from there.

ResJudicator.7916:

I personally would rather see a slightly modified (smarter) S2 system with the following changes:

• Increase RD (rating deviation) slightly for purposes of forming a team —* This gives “average” players a higher chance to team with “above average” players, or “below average” a higher chance to team with “average.” So having a high MMR is less a guarantee of victory, and a low MMR less a guarantee of loss.

• Tie MMR loss to Pip Loss —* The only regular complaint I see with S2 is about people claiming that they’ve fallen into “MMR hell”. This only happens if you’re constantly losing games (so losing MMR) without losing pips (so not dropping down and playing with worse players). This could easily be resolved by adjusting the system so that you only lose MMR when you lose a pip. In other words, if you are 0 pips into Ruby, further losses won’t decrease your MMR. Voila, no more MMR hell.

• Smarter MMR system —* The amount of MMR you gain/lose after a game should be affected by the final score. For example, the MMR impact for a 50-500 loss vs an equal-MMR team should be greater than the impact if you lost by only 495-500 to that same team. And the system could even increase your MMR if you had a close loss vs a much stronger team. This may incentivize players to try harder even when they think the loss is inevitable. (This ties in with the next suggestion)

• More transparent MMRs* — Let players see their personal MMR. Also, display the average MMRs of each team after a game. This way, players can track their own growth (or regression), and also better understand how their games were lined up. This would be particularly helpful for players who incorrectly feel like they’re stuck in MMR Hell. (For example, several of the MMR Hell gameplay vids that have been posted show that the poster was actually on the stronger team, but the poster simply lacked the awareness to realize it).

I’ve seen, even suggested myself many of these tweaks to matchmaking, and the end result would be the same.

I think the Season1 matchmaker was nearly spot on, besides the high rating deviation between players on each team, as long as that could get tweaked, bring it on. In my personal opinion, nearly all the problems with Season1 stem from the league system itself, not the matchmaker.

A tight grouping or small rating deviation inside a team is a great thing, unless of course you are up against other tightly grouped teams of random skill levels. I’m not fussed whether the rating deviation is relaxed in the way you suggested, or the same result in any other way.

Yes, league =/= skill, but not because of mmr streaks. It is because matchmaking errors in antes 50/50 loss system. The games says, oh this person has too high a win rate, I will only match him with the best players in this tier I can find to lower his win rate and mmr because he will start losing games. If you lose lots of games it will say, oh let’s make , this person get on this poor good high mmr players team to average out our win rates. And random streaks of being carried get players that can be replaced by more useful bots to diamond.

Dahkeus.8243:

Of course there’s a correlation of skill. It’s not perfect, but there’s a reason that pro league players are all legendary.

It’s also the same reason why you can watch low tier games and see huge mistakes. I mean, I steamrolled through amber, sapphire, and even most of ruby doing things like winning 1 v 1s on a thief and capping an undefended mid point. There’s still plenty of dumb stuff that happens, even in legendary games, but not nearly to the same extent that you see at lower tier games.

Not all pro players are legendary. But yes, most pro players spend more time in the game and therefore have had enough time / such above average skill or team composition that they where able to beat the carry system. Most legend players get there through trio or duo question because they are the best I’m terms of match making and allowing you to have consistent and synergistic team mates that can rotate appropriately. Some also simply farm all the way through with full premade of prologue players at off hours so they never fight any good like the people in. Division 12 legendary etc…

Chaith.8256:

ResJudicator.7916:

I personally would rather see a slightly modified (smarter) S2 system with the following changes:

• Increase RD (rating deviation) slightly for purposes of forming a team —* This gives “average” players a higher chance to team with “above average” players, or “below average” a higher chance to team with “average.” So having a high MMR is less a guarantee of victory, and a low MMR less a guarantee of loss.

• Tie MMR loss to Pip Loss —* The only regular complaint I see with S2 is about people claiming that they’ve fallen into “MMR hell”. This only happens if you’re constantly losing games (so losing MMR) without losing pips (so not dropping down and playing with worse players). This could easily be resolved by adjusting the system so that you only lose MMR when you lose a pip. In other words, if you are 0 pips into Ruby, further losses won’t decrease your MMR. Voila, no more MMR hell.

• Smarter MMR system —* The amount of MMR you gain/lose after a game should be affected by the final score. For example, the MMR impact for a 50-500 loss vs an equal-MMR team should be greater than the impact if you lost by only 495-500 to that same team. And the system could even increase your MMR if you had a close loss vs a much stronger team. This may incentivize players to try harder even when they think the loss is inevitable. (This ties in with the next suggestion)

• More transparent MMRs* — Let players see their personal MMR. Also, display the average MMRs of each team after a game. This way, players can track their own growth (or regression), and also better understand how their games were lined up. This would be particularly helpful for players who incorrectly feel like they’re stuck in MMR Hell. (For example, several of the MMR Hell gameplay vids that have been posted show that the poster was actually on the stronger team, but the poster simply lacked the awareness to realize it).

I’ve seen, even suggested myself many of these tweaks to matchmaking, and the end result would be the same.

I think the Season1 matchmaker was nearly spot on, besides the high rating deviation between players on each team, as long as that could get tweaked, bring it on. In my personal opinion, nearly all the problems with Season1 stem from the league system itself, not the matchmaker.

A tight grouping or small rating deviation inside a team is a great thing, unless of course you are up against other tightly grouped teams of random skill levels. I’m not fussed whether the rating deviation is relaxed in the way you suggested, or the same result in any other way.

S1 allowed low MMR players to reach legendary by beating low MMR teams. They achieved the division without having to face high MMR players.

Pyriall.1683:

Chaith.8256:

ResJudicator.7916:

I personally would rather see a slightly modified (smarter) S2 system with the following changes:

• Increase RD (rating deviation) slightly for purposes of forming a team —* This gives “average” players a higher chance to team with “above average” players, or “below average” a higher chance to team with “average.” So having a high MMR is less a guarantee of victory, and a low MMR less a guarantee of loss.

• Tie MMR loss to Pip Loss —* The only regular complaint I see with S2 is about people claiming that they’ve fallen into “MMR hell”. This only happens if you’re constantly losing games (so losing MMR) without losing pips (so not dropping down and playing with worse players). This could easily be resolved by adjusting the system so that you only lose MMR when you lose a pip. In other words, if you are 0 pips into Ruby, further losses won’t decrease your MMR. Voila, no more MMR hell.

• Smarter MMR system —* The amount of MMR you gain/lose after a game should be affected by the final score. For example, the MMR impact for a 50-500 loss vs an equal-MMR team should be greater than the impact if you lost by only 495-500 to that same team. And the system could even increase your MMR if you had a close loss vs a much stronger team. This may incentivize players to try harder even when they think the loss is inevitable. (This ties in with the next suggestion)

• More transparent MMRs* — Let players see their personal MMR. Also, display the average MMRs of each team after a game. This way, players can track their own growth (or regression), and also better understand how their games were lined up. This would be particularly helpful for players who incorrectly feel like they’re stuck in MMR Hell. (For example, several of the MMR Hell gameplay vids that have been posted show that the poster was actually on the stronger team, but the poster simply lacked the awareness to realize it).

I’ve seen, even suggested myself many of these tweaks to matchmaking, and the end result would be the same.

I think the Season1 matchmaker was nearly spot on, besides the high rating deviation between players on each team, as long as that could get tweaked, bring it on. In my personal opinion, nearly all the problems with Season1 stem from the league system itself, not the matchmaker.

A tight grouping or small rating deviation inside a team is a great thing, unless of course you are up against other tightly grouped teams of random skill levels. I’m not fussed whether the rating deviation is relaxed in the way you suggested, or the same result in any other way.

S1 allowed low MMR players to reach legendary by beating low MMR teams. They achieved the division without having to face high MMR players.

That’s my sense as well. If you only care about even, challenging matches, then S1’s system was good. But in S1, league placement had absolutely 0 correlation with skill. In fact, it was easier for bad players to climb because they’d get teamed up with very good players.

In S2, league placement still isn’t a great indicator of skill, but there’s at least some correlation. And many of the flaws can be fixed.

Novuake.2691:

Leagues still have zero correlation to skill, we all know this, but some very obvious things need to happen before leagues correlate to skill.

The reason for this issue is manifold, but the major ones are :
A. Too many checkpoint and safespots on the ladder.
B. Per player in game evaluation isn’t considered nor is there a system in place to do this.
C. Teams can often carry singular players because of lack of B.

Now these are my proposal to implement a way that leagues ACTUALLY correlate to skill, some of which will likely never happen for various reasons. But the goal of this post is to get leagues to correlate skill and this is how I would do it.

1. SoloQ/DuoQ and teamQ, this avoids people being carried. But also creates a team based competitive environment that is competitive.

2. MMR needs re-evaluation, it adjusts way to drastically based on 2 losses or 2 victories in a row. This leads to streaks that are extremely unhealthy.

3. Match contribution needs to be evaluated and reworked so that you can gain pips based on contribution. This will encourage people to play to the end. A loss and almost no contribution is a 2 pip loss. A win and major contribution should be a 3 pip gain. A loss but close score should award 1 pip at most if major contributer, 0 pips for above average and -1 for average and below.
Match contribution rework is what this entire suggestion hinges on, it needs to be done properly and tested extensively.

4. To accommodate the above you need to greatly increase the total pip requirement for rank climbing and slightly increase tier pip requirement.

5. Pre-season seeding. Using a certain amount of games + previous season MMR + rank to determine a starting league where the MAJORITY of players start in emerald, good players seed to sapphire and the best of the best seed to ruby, new and bad players seed to amber obviously.

6. Tier loss should be implemented from emerald and up. Rank loss should be implemented from ruby(so you can drop down to sapphrie, where the average players should be after a season). You could drop to a lower rank if you lose 3 consecutive games with below average contribution after reaching zero pips.

Yes, some of these are pretty unspecific, but i am terrible at match, so ill leave the to the people that know better.

Feel free to discuss and improve on my ideas. In a perfect world we might get this.

EDIT : I want to add that this system would mean you climb if you play well, instead of climbing by streaking.
So if you win 1, lost 1 constantly, which is HEALTHY, you should still climb if you played well, which is extremely important.

As long as you have sound matchmaking, that is as you said, not streaky, then yes, I can get behind things like division loss, tier loss and seeding.

If anything, division loss can really help people in lower divisions advance. As far as pips go, I think you can code for different situations. I’d say +1 pip for win(obviously), no pips loss for close defeat(490-500), -1 pip for loss > 50 pts, +2 pips for stomps(win > 200pts) etc.

This system just rewards and punishes people unnecessarily primarily based on their wins/loss streaks thanks to the matchmaking.

ResJudicator.7916:

Pyriall.1683:

Chaith.8256:

ResJudicator.7916:

I personally would rather see a slightly modified (smarter) S2 system with the following changes:

• Increase RD (rating deviation) slightly for purposes of forming a team —* This gives “average” players a higher chance to team with “above average” players, or “below average” a higher chance to team with “average.” So having a high MMR is less a guarantee of victory, and a low MMR less a guarantee of loss.

• Tie MMR loss to Pip Loss —* The only regular complaint I see with S2 is about people claiming that they’ve fallen into “MMR hell”. This only happens if you’re constantly losing games (so losing MMR) without losing pips (so not dropping down and playing with worse players). This could easily be resolved by adjusting the system so that you only lose MMR when you lose a pip. In other words, if you are 0 pips into Ruby, further losses won’t decrease your MMR. Voila, no more MMR hell.

• Smarter MMR system —* The amount of MMR you gain/lose after a game should be affected by the final score. For example, the MMR impact for a 50-500 loss vs an equal-MMR team should be greater than the impact if you lost by only 495-500 to that same team. And the system could even increase your MMR if you had a close loss vs a much stronger team. This may incentivize players to try harder even when they think the loss is inevitable. (This ties in with the next suggestion)

• More transparent MMRs* — Let players see their personal MMR. Also, display the average MMRs of each team after a game. This way, players can track their own growth (or regression), and also better understand how their games were lined up. This would be particularly helpful for players who incorrectly feel like they’re stuck in MMR Hell. (For example, several of the MMR Hell gameplay vids that have been posted show that the poster was actually on the stronger team, but the poster simply lacked the awareness to realize it).

I’ve seen, even suggested myself many of these tweaks to matchmaking, and the end result would be the same.

I think the Season1 matchmaker was nearly spot on, besides the high rating deviation between players on each team, as long as that could get tweaked, bring it on. In my personal opinion, nearly all the problems with Season1 stem from the league system itself, not the matchmaker.

A tight grouping or small rating deviation inside a team is a great thing, unless of course you are up against other tightly grouped teams of random skill levels. I’m not fussed whether the rating deviation is relaxed in the way you suggested, or the same result in any other way.

S1 allowed low MMR players to reach legendary by beating low MMR teams. They achieved the division without having to face high MMR players.

That’s my sense as well. If you only care about even, challenging matches, then S1’s system was good. But in S1, league placement had absolutely 0 correlation with skill. In fact, it was easier for bad players to climb because they’d get teamed up with very good players.

In S2, league placement still isn’t a great indicator of skill, but there’s at least some correlation. And many of the flaws can be fixed.

I disagree, the corrolation was greater or if not better last season. Why? How do you explain the fact that most S1 diamond got to legendary this season? True they might have improved, but ask anyone S1 legendary, it will tell you their game quality are worse. If that’s not an indication of the skill level of most players on legendary then idk what is.

And about folks who climbed to legendary by playing against other MMR players, I am having trouble believing this. Simply because, last season you had to have at 55 60 % winrate to progress. So by the time, the " low MMR player" get to diamond, his MMR would pretty much be stable and within range of other diamond players. So how is possible for someone with kitten to 60% winrate to keep on facing low MMR players?

Also if I may, last season pro players complained about have low MMR on their team, so my semblance of S1 MMR is 3 2 3 2 1 vs 4 1 1 4 2. Still, I acknowledge that low MMR vs Low MMR happen but the trend stopped in diamond. Because you don’t get to win 30 games ( amount of games needed to get to diamond) without it affect your MMR by at least ( 30*30 = 900 pts). Granted you can win one and lose another, in the end you still needed ( in s1) a total of 30 wins (net value) to progress. Conceptually, IA ma struggling with your assertion of low MMR legendary.

And also, most folks got carried to legendary last season ( pro players included) by using GOD DRUIDS, TANKIESTEST ELE or bunker mesmer. And let’s not forget that those classes carry potential were better than S2 reaper’s. Now are we going to blame this on the meta or the algorithm?

Chaith.8256:

So the goal is to give additional pips to individual players based on merit, but I’m not convinced that is necessary to having a skill based league system. It might even create problems, like division getting way too high for their skill, via boosted rewards for beating on minionmancers & eles all game and accomplishing nothing.

I have a rather simple proposal to change the matchmaker so this problem doesn’t happen in the first place, and that’s simply ship a priority change of how MMR grouping works.

In any pip range we have a variety of MMRs, let’s group them into 3 camps – below average, average, and above average.

Right now, from my understanding of the matchmaker, we’ll have a team of 5 taken from the below average camp and randomly faced off against a team from the below average, average, or above average camp.

When ArenaNet heard our desires of having teammates relatively close to us in skill, this MMR grouping was absolutely NOT what we wanted, if I can be so bold as to speak for the general population.

The matchmaking in season1 was good in the way that it had high quality, close matches. A big problem (excluding other problems) is that it did whatever it took to even the team MMRs, including adding players who didn’t belong there, to handicap, or boost the team MMR. This was disruptive.

What I propose is a combination of S1 and S2 matchmaking, where matchmaking still determines your matchups based on pip range, like it does now, to stay in line of ‘perserving prestige’ in divisions.

The main difference would be instead of finding two very different groups of 5 alike MMR people, and having a potentially imbalanced match, try to find the 10 most similarly rated players in MMR within the pip range, rank them #1-10 based on MMR, and have even vs. odds. With an expanding PIP range, and variation in MMRs accepted as the queue goes on.

In a 5v5 with the 10 most similarly rated players, there won’t be as big a need to promote certain players over others, as it’ll be much harder for you to become underrated in MMR by the system.
[…]
I’m a bit weary of all the crazy matchmaker/ladder fads we’ve tested. I’m happy we added the league system which was a huge step forward, but we just need to get a normal and ordinary matchmaker to complement it like we had before.

I also support this idea.

Fivedawgs.4267:

Nah, slightly modified in the wrong way?? Just give people MMR vs MMR or pip vs pip range. That’s what lol, data or Smite or CS:GO uses. So what aren’t we doing that?? Even the Glicko 2 excel add on won’t let me match a 1000 vs a 3000 chess player.

Start with the basic MMR vs MMR (within 2 tier) or pip vs pip( this one is risky though) and go from there.

I think pip v pip relies on a large population to be effective and I don’t think GW2 has that. I think there is a huge risk that we would end up in a similar situation to this season. It has to be mmr based.

Fivedawgs.4267:

And about folks who climbed to legendary by playing against other MMR players, I am having trouble believing this. Simply because, last season you had to have at 55 60 % winrate to progress. So by the time, the " low MMR player" get to diamond, his MMR would pretty much be stable and within range of other diamond players. So how is possible for someone with kitten to 60% winrate to keep on facing low MMR players?

Also if I may, last season pro players complained about have low MMR on their team, so my semblance of S1 MMR is 3 2 3 2 1 vs 4 1 1 4 2. Still, I acknowledge that low MMR vs Low MMR happen but the trend stopped in diamond. Because you don’t get to win 30 games ( amount of games needed to get to diamond) without it affect your MMR by at least ( 30*30 = 900 pts). Granted you can win one and lose another, in the end you still needed ( in s1) a total of 30 wins (net value) to progress. Conceptually, IA ma struggling with your assertion of low MMR legendary.

There’s a few really easy ways low MMR players could reach legendary in Season 1 without ever playing high-MMR players:

Method 1: If you start with low MMR, just winning/losing at about a 50% ratio would slowly get you to legendary without raising your MMR. This is because you could gain pips from losing to a team with higher MMR. In fact, it was possible to reach legendary with less than a 50% win ratio for this reason.

Method 2: Even if you somehow started with an inflated MMR, you could tank it at a safe zone (i.e. area where there are no pip losses) by losing repeatedly. Then climb to the next safe zone courtesy of getting easy games due to your now-low MMR. Then tank again from losing repeatedly.

And even if you were queued into games with proleague players (due to low population numbers), your low MMR would put them onto your team. That’s what we mean by getting to legendary in Season 1 without having to face high-MMR players. The only time you’d ever have to face a high-MMR player in Season 1 is if you queued at an off-hour.

@Fivedawgs.4267

Winrate doesn’t correlate to progression if there are so many checkpoints.

ResJudicator.7916:

Fivedawgs.4267:

And about folks who climbed to legendary by playing against other MMR players, I am having trouble believing this. Simply because, last season you had to have at 55 60 % winrate to progress. So by the time, the " low MMR player" get to diamond, his MMR would pretty much be stable and within range of other diamond players. So how is possible for someone with kitten to 60% winrate to keep on facing low MMR players?

Also if I may, last season pro players complained about have low MMR on their team, so my semblance of S1 MMR is 3 2 3 2 1 vs 4 1 1 4 2. Still, I acknowledge that low MMR vs Low MMR happen but the trend stopped in diamond. Because you don’t get to win 30 games ( amount of games needed to get to diamond) without it affect your MMR by at least ( 30*30 = 900 pts). Granted you can win one and lose another, in the end you still needed ( in s1) a total of 30 wins (net value) to progress. Conceptually, IA ma struggling with your assertion of low MMR legendary.

There’s a few really easy ways low MMR players could reach legendary in Season 1 without ever playing high-MMR players:

Method 1: If you start with low MMR, just winning/losing at about a 50% ratio would slowly get you to legendary without raising your MMR. This is because you could gain pips from losing to a team with higher MMR. In fact, it was possible to reach legendary with less than a 50% win ratio for this reason.

Method 2: Even if you somehow started with an inflated MMR, you could tank it at a safe zone (i.e. area where there are no pip losses) by losing repeatedly. Then climb to the next safe zone courtesy of getting easy games due to your now-low MMR. Then tank again from losing repeatedly.

And even if you were queued into games with proleague players (due to low population numbers), your low MMR would put them onto your team. That’s what we mean by getting to legendary in Season 1 without having to face high-MMR players. The only time you’d ever have to face a high-MMR player in Season 1 is if you queued at an off-hour.

I agree with you both method however I disagree with your assertion that low MMR players were the ones reaping the benefits of the system( that’s is exploiting it to get to leendary.) The way I see it, average and high tier player were the ones who exploited it not a low MMR player. And here is why:

First let me start by making the following assumption

-As it stand, the lowest rating is 100 . The default 1500 and the highest 5000. OK let’s assume that most players are around the default rating; and with 350 being the maximal deviation. Most average players are between 1150 – 1850 and anyone who is below 1150 is a low MMR player and one above 1850 is a high MMR player.

-Second point, the S1 meta aside, let’s look at the player potential in their respective MMR zone; and for the sake of time, I’ll mainly focus on low MMR players.

A low MMR player is most likely learning the game, still learning his rotation, have a basic knowledge of who does what, still learning his counters .etc… my point being he knows he can’t carry( the whole notion o carrying in 5 vs 5 game is absurd but it does happen) a game.

-finally, I want to put it out there, MMR != SKILL, However most high MMR players are skilled, likewise most Low MMR players are unskilled.

And wit out more ado,

Method 1 : in short, you are saying you could get to leendary (even with a 50% winrate) simply because you can gain pip by losing. True, but out of those 30 pip how many were gained by losing, since you said the player has a 50% winrate until ruby. OK, let’s assume that the player MMR upon crossing to ruby is (100 + 1150/2 = 625). The only way this player would keep on facing low MMR player in diamond is if 50 % of his 30 pips( Net) needed to advanced came from losing that’s is ( 15 × 30 = 450 .. Player Rating in diamond = 625 + 450 = 1075 which is still low) and that’s quite hard to accomplish ( that’s is progressing by losing ) or if like you said (method 2) upon getting to diamond the Low MMR player with a now Average MMR ( since gaining 15 pip by losing is almost impossible) tanked his MMR again.

Based on the assumptions above, a Low MMR player confidence is not high enough that he would be willing to lose a few games in hope of getting easier matches or winning the next one. Why because he is still learning the game, etc… he has close to zero carry potential (meta bunker aside; using carry for the sake of argument). But you see where I am going, to throw games( tank your MMR) you have to be sure/certain that you can carry the next one. Most low MMR players just don’t have that mindset.

It seems as if average or high MMR players were the one who got to legendary by first lowering their MMR to avoid facing their peers; because they knew they would win their next game.

I can see how it can happen, but it’s synonymous to gambling your house to a low MMR players because you he can’t garantee the outcome of his game. All this assuming that the system only did 1 1 1 1 1 vs 1111 or 5 5 5 5 5 vs 5 5 5 5 5; because my anecdotal S1 experience differs from that as I had a mix of players( went up against pro players in ruby and had some in my team last season or simply 3 2 2 4 1 vs 4 1 1 4 2).

Conclusion: it’s highly unlikely, for a low MMR player ( low MMR player advancing by losing or tanking his MMR even though he can’t garantee his next win) to luck ride his way to.legendary with the S1 MM algorithm, bunker meta aside.

Fivedawgs.4267:

ResJudicator.7916:

Fivedawgs.4267:

And about folks who climbed to legendary by playing against other MMR players, I am having trouble believing this. Simply because, last season you had to have at 55 60 % winrate to progress. So by the time, the " low MMR player" get to diamond, his MMR would pretty much be stable and within range of other diamond players. So how is possible for someone with kitten to 60% winrate to keep on facing low MMR players?

Also if I may, last season pro players complained about have low MMR on their team, so my semblance of S1 MMR is 3 2 3 2 1 vs 4 1 1 4 2. Still, I acknowledge that low MMR vs Low MMR happen but the trend stopped in diamond. Because you don’t get to win 30 games ( amount of games needed to get to diamond) without it affect your MMR by at least ( 30*30 = 900 pts). Granted you can win one and lose another, in the end you still needed ( in s1) a total of 30 wins (net value) to progress. Conceptually, IA ma struggling with your assertion of low MMR legendary.

There’s a few really easy ways low MMR players could reach legendary in Season 1 without ever playing high-MMR players:

Method 1: If you start with low MMR, just winning/losing at about a 50% ratio would slowly get you to legendary without raising your MMR. This is because you could gain pips from losing to a team with higher MMR. In fact, it was possible to reach legendary with less than a 50% win ratio for this reason.

Method 2: Even if you somehow started with an inflated MMR, you could tank it at a safe zone (i.e. area where there are no pip losses) by losing repeatedly. Then climb to the next safe zone courtesy of getting easy games due to your now-low MMR. Then tank again from losing repeatedly.

And even if you were queued into games with proleague players (due to low population numbers), your low MMR would put them onto your team. That’s what we mean by getting to legendary in Season 1 without having to face high-MMR players. The only time you’d ever have to face a high-MMR player in Season 1 is if you queued at an off-hour.

I agree with you both method however I disagree with your assertion that low MMR players were the ones reaping the benefits of the system( that’s is exploiting it to get to leendary.) The way I see it, average and high tier player were the ones who exploited it not a low MMR player. And here is why:

First let me start by making the following assumption

-As it stand, the lowest rating is 100 . The default 1500 and the highest 5000. OK let’s assume that most players are around the default rating; and with 350 being the maximal deviation. Most average players are between 1150 – 1850 and anyone who is below 1150 is a low MMR player and one above 1850 is a high MMR player.

-Second point, the S1 meta aside, let’s look at the player potential in their respective MMR zone; and for the sake of time, I’ll mainly focus on low MMR players.

A low MMR player is most likely learning the game, still learning his rotation, have a basic knowledge of who does what, still learning his counters .etc… my point being he knows he can’t carry( the whole notion o carrying in 5 vs 5 game is absurd but it does happen) a game.

-finally, I want to put it out there, MMR != SKILL, However most high MMR players are skilled, likewise most Low MMR players are unskilled.

And wit out more ado,

Method 1 : in short, you are saying you could get to leendary (even with a 50% winrate) simply because you can gain pip by losing. True, but out of those 30 pip how many were gained by losing, since you said the player has a 50% winrate until ruby. OK, let’s assume that the player MMR upon crossing to ruby is (100 + 1150/2 = 625). The only way this player would keep on facing low MMR player in diamond is if 50 % of his 30 pips( Net) needed to advanced came from losing that’s is ( 15 × 30 = 450 .. Player Rating in diamond = 625 + 450 = 1075 which is still low) and that’s quite hard to accomplish ( that’s is progressing by losing ) or if like you said (method 2) upon getting to diamond the Low MMR player with a now Average MMR ( since gaining 15 pip by losing is almost impossible) tanked his MMR again.

Based on the assumptions above, a Low MMR player confidence is not high enough that he would be willing to lose a few games in hope of getting easier matches or winning the next one. Why because he is still learning the game, etc… he has close to zero carry potential (meta bunker aside; using carry for the sake of argument). But you see where I am going, to throw games( tank your MMR) you have to be sure/certain that you can carry the next one. Most low MMR players just don’t have that mindset.

It seems as if average or high MMR players were the one who got to legendary by first lowering their MMR to avoid facing their peers; because they knew they would win their next game.

I can see how it can happen, but it’s synonymous to gambling your house to a low MMR players because you he can’t garantee the outcome of his game. All this assuming that the system only did 1 1 1 1 1 vs 1111 or 5 5 5 5 5 vs 5 5 5 5 5; because my anecdotal S1 experience differs from that as I had a mix of players( went up against pro players in ruby and had some in my team last season or simply 3 2 2 4 1 vs 4 1 1 4 2).

Conclusion: it’s highly unlikely, for a low MMR player ( low MMR player advancing by losing or tanking his MMR even though he can’t garantee his next win) to luck ride his way to.legendary with the S1 MM algorithm, bunker meta aside.

What? I think you’re overcomplicating things without adding any extra accuracy.

Here’s the easiest, cleanest way I can explain it: In S1, every win gives you at least 1 pip (up to 3 pips). Losses can award 1 pip, 0 pips, or subtract 1 pip (depending on MMR difference). So with a 50% winrate, you have a greater-than-50% chance of gaining a pip every game. This means that a 50% winrate will slowly push you up the leaderboard, with very little MMR growth.

So which MM algorithm did S1 use? MMR vs MMR ? If yes then only one pip was awarded per wins and few were those who gained 2 or 3 pip for losing or winning.

And my understanding of the extra pip is when one of the player on your team was a pro or whatever. But most people would agree it was MMR vs MMR; knowing how glicko 2 works, I can say with 95% certainty that less than 1% of the player base gained more than 10 pip total (extra from win and 5hose for loss) last season.

Finally, I am not disputing the growth because once there is a possibility anything can happen. What I am disputing is YOUR ASSEMENT THAT LOW MMR PLAYERS used method 1 and 2 to get to legendary, and I have pretty much explained why ( bunker meta aside.)

Fivedawgs.4267:

So which MM algorithm did S1 use? MMR vs MMR ? If yes then only one pip was awarded per wins and few were those who gained 2 or 3 pip for losing or winning.

And my understanding of the extra pip is when one of the player on your team was a pro or whatever. But most people would agree it was MMR vs MMR; knowing how glicko 2 works, I can say with 95% certainty that less than 1% of the player base gained more than 10 pip total (extra from win and 5hose for loss) last season.

Finally, I am not disputing the growth because once there is a possibility anything can happen. What I am disputing is YOUR ASSEMENT THAT LOW MMR PLAYERS used method 1 and 2 to get to legendary, and I have pretty much explained why ( bunker meta aside.)

But you haven’t explained it. You just made assumptions that make no sense. Your purported knowledge of Glicko-2 is completely irrelevant — what you need is data about the MMR distribution curve in S1, along with player activity.

You apparently agree that anyone could reach Legendary in S1 with a 50% winrate. So now you argue that only ~1% of people did so — a statistic you’ve completely made up without any data. I don’t know how many “low MMR” people hit Legendary, either — only A.net has that answer. But it was obviously a significant enough number to generate all the “I made Legendary in S1 but am in MMR hell for S2” posts that we now see.

ResJudicator.7916:

Fivedawgs.4267:

So which MM algorithm did S1 use? MMR vs MMR ? If yes then only one pip was awarded per wins and few were those who gained 2 or 3 pip for losing or winning.

And my understanding of the extra pip is when one of the player on your team was a pro or whatever. But most people would agree it was MMR vs MMR; knowing how glicko 2 works, I can say with 95% certainty that less than 1% of the player base gained more than 10 pip total (extra from win and 5hose for loss) last season.

Finally, I am not disputing the growth because once there is a possibility anything can happen. What I am disputing is YOUR ASSEMENT THAT LOW MMR PLAYERS used method 1 and 2 to get to legendary, and I have pretty much explained why ( bunker meta aside.)

But you haven’t explained it. You just made assumptions that make *no sense. *

A Show me where my logic is faulty so i can beyter it for you.

Your purported knowledge of Glicko-2 is completely irrelevant — what you need is data about the MMR distribution curve in S1, along with player activity.

A: I am using a normally distributed curve which is pretty standard for humans studies. Since that’s the tone you want to use, any of your above post are hot air as well. Unless, you managed to get your hand on the data.

You apparently agree that anyone could reach Legendary in S1 with a 50% winrate.

A: i guess by saying possible. I am agreeing with you.

So now you argue that only ~1% of people did so — a statistic you’ve completely made up without any data.

A: Nope, i never said this, you read my wrong. I agree witht eh 50% winrate tp legendary. Because a pro player last season could tank his MMR as low as he wanted; due to the simple fact HE COULD CARRY his next game. A low MMR player cant,. Futhermore , I said less than 1% of the player base gained more than 10 pips total last season by either losing or accumulating extra pip. So, the percentage have nothing to do with the 50% winrate to legendary.I have used it to debunked the Low MMR vs Low MMR in diamond

I don’t know how many “low MMR” people hit Legendary, either — only A.net has that answer. But it was obviously a significant enough number to generate all the “I made Legendary in S1 but am in MMR hell for S2” posts that we now see.

_A: don’t they have a point? Seing how the system works, i woukdnt be surprised.First. I have shown to you to the best of my abilities that a low MMR couldn’t get to elegndary by using any of your methods (bunker meta aside). Now, the meta could have carried a lot MMR player to legendary just like reapers and the system are carrying some this season. Now, if I may inquire about those S1 legendary players , where are they now?? Haven’t they all reached legendary by now? (Maybe mot as fast ad they wanted) Just curious.

And if you can show me how a LOW MMR player could have got to carried to legendary by the system (not the meta) last season. Since you are so hell bent on the idea that lots of them did.

Fivedawgs.4267:

Words

OK, I think I see where your confusion comes from. In S1, you didn’t have to carry games if you had a low MMR. Having a low MMR meant that (1) You were more likely to get high MMR players on your team; and (2) You were more likely to get low MMR players on the enemy team.

In other words, tanking your MMR helped you get carried to legend.

In S2, it’s the opposite: having a low MMR now makes it very difficult to advance.

ResJudicator.7916:

Fivedawgs.4267:

Words

OK, I think I see where your confusion comes from. In S1, you didn’t have to carry games if you had a low MMR. Having a low MMR meant that (1) You were more likely to get high MMR players on your team; and (2) You were more likely to get low MMR players on the enemy team.

In other words, tanking your MMR helped you get carried to legend.

In S2, it’s the opposite: having a low MMR now makes it very difficult to advance.

Accurate, but not the only issue this season has