Guild Wars 2

Maybe this has been discussed previously but the current stagnation of several tiers in NA got me looking into the rating system ANet uses. It seems that many servers are “stuck” in tiers they really should not be in. I am on SoR and we have been fighting BG since Nov 17th and have lost each match yet they have not gained enough rating in this time to move to T1 even though they clearly are not really challenged in T2. Also during this time we (SoR and BG and SoS a bit too) wiped the floor with IoJ and they remained in T2 for far too long. They have since moved to T3 and seem very competitive in that tier.

The thread title is a bit misleading as I do not think that the Glicko 2 rating system is really to blame for these stagnant matches. I believe it is the score differential equations ANet uses to input into the Glicko system.
(See: http://www.guildwars2guru.com/arenanet-tracker/topic/244038-the-math-behind-wvw-ratings/)

They utilize an equation (wAGlickoScore = (sin((wAPercent – 0.5) * Pi) + 1) * 0.5)
that essentially decreases the effective rating changes for very large blow outs (never mind that this equation will never give a 0 to 1 result that is required for Glicko, but I digress), thereby keeping servers in a tier longer than their performance would dictate. I believe that this should be the opposite perhaps the following equation graph

The largest realistic relative score difference are going to be in the range of 0.3 to 0.6 for blow-outs between the bottom and top server. This is really the region of interest in any Glicko input function. The graph that I linked will have a slight non-linear increase in rating for these blowouts for the top server. This would allow for rivalries to form but might cut down on the stagnation we are seeing.

These are just the musings of a scientist on vacation so any feedback or input is appreciated

TLDR: blame anet and their modification of the Glicko not Glicko for current rating stagnation

Wrong science – it is more a function of the population and human nature than mathematics.

Free server transfers combined with the natural tendency of most people to transfer to the top X percent servers has created an segmented landscape that continually pushes the larger numbers to a smaller and smaller number of servers – which creates buckets based on population/coverage that manifests as static/stagnant rankings.

While skill and coordination can play a role, most will agree that WvW matches are usually determined by sheer numbers and how much coverage you can spread over a weeks time.

In short, the population created this issue by top loading servers. If people want to “fix” the issue – the answer is simple — be open to moving (or better yet, moving your guild) to a lower ranked server.

agreed that server xfers are a problem and free transfers should be closed but that still doesn’t explain the fact that servers are not changing ranks even with “fair weather” transfers. Winning servers seem to get the bulk of these xfers and they still struggle to gain rank. I am simply trying to find a solution that would be easily implemented that would lessen the current tier stagnation.

Ratings should be reset every 6 months, like a tournament system. Mix the servers up between win/lose/middle across the tiers. Give everyone a fair chance to “reset” where their server sits in the tiers and re-rate. When 6 months is up, do it all over again. Or, find another system that matches WvW populations together and gives them some chance to fight against other similar WvW populations. Can’t use server pops because it’d be possible to find servers that primarily PvE and that would of course mess up the ratings.

OR, some other system that doesn’t allow a losing server to basically not play at all and lose and still be in 1st place in their tier (which I’ve seen often happen). THAT is not a rating system…THAT’s a JOKE.

The system is flawed, thanks for your post.

The OP raises some good points about the implementation of the Glicko system.

Blaeys.3102:

While skill and coordination can play a role, most will agree that WvW matches are usually determined by sheer numbers and how much coverage you can spread over a weeks time.

Not anyone who plays WvW more than casually and doesn’t get caught up in making excuses for their losses.

Tell that to JQ who is about to kick us out of first place, not because they have coverage advantage, but because they played better than us this week. JQ, SBI, BG, and SoS all have roughly the same amount of coverage, the winners are determined by who plays better.

The reality is that every server had coverage gaps, but it’s about how hard you play during primetimes that really matters. SoS has a huge coverage gap between NA and Oceanic, but we’ve been #1 for over a month because we’ve owned the PPT when all 3 servers are out in full force with nearly identical numbers.

Alarox.4590:

The OP raises some good points about the implementation of the Glicko system.

Blaeys.3102:

While skill and coordination can play a role, most will agree that WvW matches are usually determined by sheer numbers and how much coverage you can spread over a weeks time.

Not anyone who plays WvW more than casually and doesn’t get caught up in making excuses for their losses.

Tell that to JQ who is about to kick us out of first place, not because they have coverage advantage, but because they played better than us this week. JQ, SBI, BG, and SoS all have roughly the same amount of coverage, the winners are determined by who plays better.

Dont be insulting and keep the discussion meaningful. It isnt about excuses.

It’s why I used words like usually. Yes, they may be knocking you out of first place, but the brackets are still staying rather static – and the reason for that is that the three or four top populated servers as a whole draw in more people across more times of the day. When compared to servers further down, the first – and MOST IMPORTANT – difference is numbers – its impossible to know where the better players are because the severe number imbalance trumps everything else. Ive played on Henge of Denravi since the first beta weekend in May, so Ive seen this from both sides of the coin.

The original topic was about what caused the problem of static brackets. My opinion – it is a self perpetuating vicious cycle that feeds the higher population servers and widens the divide significantly as time progresses. It is just basic human nature.

The glicko system would work if a larger number of servers were more even. Then, the differences between servers WOULD be based on skill and organization. That is what we all should be striving for.

Blaeys.3102:

Alarox.4590:

The OP raises some good points about the implementation of the Glicko system.

Blaeys.3102:

While skill and coordination can play a role, most will agree that WvW matches are usually determined by sheer numbers and how much coverage you can spread over a weeks time.

Not anyone who plays WvW more than casually and doesn’t get caught up in making excuses for their losses.

Tell that to JQ who is about to kick us out of first place, not because they have coverage advantage, but because they played better than us this week. JQ, SBI, BG, and SoS all have roughly the same amount of coverage, the winners are determined by who plays better.

Dont be insulting and keep the discussion meaningful. It isnt about excuses.

It’s why I used words like usually. Yes, they may be knocking you out of first place, but the brackets are still staying rather static – and the reason for that is that the three or four top populated servers as a whole draw in more people across more times of the day. When compared to servers further down, the first – and MOST IMPORTANT – difference is numbers – its impossible to know where the better players are because the severe number imbalance trumps everything else. Ive played on Henge of Denravi since the first beta weekend in May, so Ive seen this from both sides of the coin.

The original topic was about what caused the problem of static brackets. My opinion – it is a self perpetuating vicious cycle that feeds the higher population servers and widens the divide significantly as time progresses. It is just basic human nature.

The glicko system would work if a larger number of servers were more even. Then, the differences between servers WOULD be based on skill and organization. That is what we all should be striving for.

1.) My post was never meant to be insulting, as yours wasn’t either. However, the line of your post is basically an insult to anyone who believes that their server earns their wins. No hard feelings over anything, just pointing it out.

2.) As for the supposed issue of population imbalance, the idea of spreading all of the players out is more or less spreading the misery. Unless everyone is happy, nobody can be happy. That isn’t your intention in suggesting it, but I believe that’s the reality of it. In the higher tier servers, your server actually earns the win rather than coverage or other factors being out of the players control. That’s why people that truly love WvW flock to the higher tiers. Spreading all of the players out means everyone suffers from the coverage gaps and the lack of constant competition, rather than just the lower tier servers.

At the moment, the higher tier servers are generally more of what WvW players want. At the moment, the higher tier servers are also the ones that have most of the WvW population. Therefore, at the moment, most of the WvW population is in a situation where it’s what they want. Spreading them out generally harms most of the WvW players rather than helping them. Until someone can prove the above to be false, I see no reason why spreading the WvW players out is beneficial to the WvW community at large.

Alarox.4590:

Blaeys.3102:

Alarox.4590:

The OP raises some good points about the implementation of the Glicko system.

Blaeys.3102:

While skill and coordination can play a role, most will agree that WvW matches are usually determined by sheer numbers and how much coverage you can spread over a weeks time.

Not anyone who plays WvW more than casually and doesn’t get caught up in making excuses for their losses.

Tell that to JQ who is about to kick us out of first place, not because they have coverage advantage, but because they played better than us this week. JQ, SBI, BG, and SoS all have roughly the same amount of coverage, the winners are determined by who plays better.

Dont be insulting and keep the discussion meaningful. It isnt about excuses.

It’s why I used words like usually. Yes, they may be knocking you out of first place, but the brackets are still staying rather static – and the reason for that is that the three or four top populated servers as a whole draw in more people across more times of the day. When compared to servers further down, the first – and MOST IMPORTANT – difference is numbers – its impossible to know where the better players are because the severe number imbalance trumps everything else. Ive played on Henge of Denravi since the first beta weekend in May, so Ive seen this from both sides of the coin.

The original topic was about what caused the problem of static brackets. My opinion – it is a self perpetuating vicious cycle that feeds the higher population servers and widens the divide significantly as time progresses. It is just basic human nature.

The glicko system would work if a larger number of servers were more even. Then, the differences between servers WOULD be based on skill and organization. That is what we all should be striving for.

1.) My post was never meant to be insulting, as yours wasn’t either. However, the line of your post is basically an insult to anyone who believes that their server earns their wins. No hard feelings over anything, just pointing it out.

2.) As for the supposed issue of population imbalance, the idea of spreading all of the players out is more or less spreading the misery. Unless everyone is happy, nobody can be happy. That isn’t your intention in suggesting it, but I believe that’s the reality of it. In the higher tier servers, your server actually earns the win rather than coverage or other factors being out of the players control. That’s why people that truly love WvW flock to the higher tiers. Spreading all of the players out means everyone suffers from the coverage gaps and the lack of constant competition, rather than just the lower tier servers.

At the moment, the higher tier servers are generally more of what WvW players want. At the moment, the higher tier servers are also the ones that have most of the WvW population. Therefore, at the moment, most of the WvW population is in a situation where it’s what they want. Spreading them out generally harms most of the WvW players rather than helping them.

My only intention was to explain why the brackets are as static as they are. Changing the ranking system so that teams moved brackets more than they do would only serve to create mismatched fights, which no one would enjoy.

the population issue is systemic – infecting both WvW and PVE – and not one that can simply be pushed aside (which, I dont think you are doing, but it still needed saying). People on all servers should have access to a fun experience – not just the one’s lucky (?) enough to cram into the golden 3 at the top.

Outside of server merges, not sure what the solution to that particular problem is – I just believe that the ranking system isnt to blame for the issue at hand – good or bad, it comes back to numbers. Until that issue is addressed, anything else will just create more painful scenarios, not better ones.

It’s amusing: Shortly after launch, everyone complained about the horrible mismatches and how much movement there was between brackets. The entire point of the rating system was to eventually settle on stable brackets. Now that it’s done that, people are complaining about the lack of inter-bracket volatility.

There were those that cautioned people to be patient, give the system time to work, and they’d see that eventually the system would settle into a stable state, with the same matchups week after week. Many didn’t believe it’d happen.

Well, it did happen. Just as it was supposed to. Why is this a problem? In any bracket, there will always be two losers. In a system of stabilized matches, chances are it’ll be the same two losers every match, until the winner gets worse or one of the losers gets better.

That’s hardly a surprise.

It is not really a problem but the current system keeps servers in a tier far longer that they should be. That is evident over and over again; SoS was in T2 forever then when the finally moved to T1 they did very well, same goes for IoJ moving from T2 to T3, now they look really good.

The original point of the post was not to re-hash the issue of server coverage or population it was to call attention to how the current ranking system operates and how it is flawed.

You win you move up, you lose you move down.

Whats so difficult about having that system than this glicko thing?

Ohai.8346:

You win you move up, you lose you move down.

Whats so difficult about having that system than this glicko thing?

Because you have to take into account what the people above and below you do, and how well and how often they’ve done it. Not to mention how well and how often your server’s won.

And how often and how badly each server loses.

That’s why it’s a points-based rating system. It was designed to stabilize. If you want to move to another bracket, win often enough by enough points that you beat the server above you (and hope they have a horrible losing streak). If you want to move down, lose abysmally, consistently, until your score is lower than the server below you, and hope they do really well.

The problem with moving up or down one spot based on win or loss is that it doesn’t allow for quick movement across multiple slots. And it’ll allow for too much volatility due to the nature of the game. You’d always have servers changing brackets in positions 1 and 3 of each bracket after things “settled” in that system (for a very loose value of settling, since a system like that never really does).

Well it should not difficult but the system they implemented does not operate this way. Glicko uses past performance to calculate rating (among other things) which biases well performing servers down and match upsets up. Also Anet added their own factors that decrease rating gains from dominant servers in a tier.