Guild Wars 2

Disclaimer: The first comment is for the players still struggling to understand the system, the second comment is directed towards the developers. (I forgot btw that Anet already plans to implement leagues and ladders)

The leaderboards have been an issue for most for a long time, I presume.
They came rather late to the party and were also not gently welcomed by the time frustration had built up to the release of them.
I want to propose some review of the pros and cons of the current system and possible solutions. For more information visit the glicko 1 or 2 entry on wiki and/or the professor’s own homepage.

1. Rating deviation
Rating deviation is very high at start (I assume 350 with a start value for the rating of 1500) and will go lower by the matches or time played.
This means that the more a players has played, the lower the rating deviation is and the smaller the range of rating is for opponents the player will face.
This means, the more one plays, the more they keep a constant rating.
The formula for it is r-2*RD for lower interval boundary and r+2*RD for higher interval boundary.
There is a % of confidence of how true this statement or therefore rating/value is.
Issue: This can be an issue for these on a losing streak and pin-downs them to a certain rating after a certain amount of matches or time.
Also, if not matches are unrated, the RD’ will change but the rating not.
Issue 2: With a high decay, there will be rating loss (I assume decay works externally as maybe a constant) and therefore one who has an unlucky streak in e.g. SoloQ will be punished no matter what his skill is.

2. Rating period
I have seen that the initial amount to get out of the unrated bracket is to play 10-15 games (a medium to large amount as was recommended).
The length for a rating period is determined by an administrator (obviously who it is here). For us, the rating period has started since the leaderboards were released.
Issue: 10-15 games are not really crucial to determine the true rating (and “skill” of a player).
Issue 2: If a decent player starts later, he is punished by having a different pool than decent players who started earlier.

3. Rating volatility
This is actually seems to have less effect in general. It describes the consistency. A low value of rating volatility describes a consistent level of performance while a high one displays a sudden peak and change in performance.
Issue: No transparency on any data for the players. While this is not much of an issue for players as they can see first-hand if they lose or win and can improve from mistakes done during the matches, it may still show interesting data.
It would involve meta changes more often for example so players can see that if they played a certain build, their performance was consistent and invovled mostly wins or losses. It would give players some more feedback.

4. System constant
The system constant restraints large volatility measure changes so the rating changes less.
Issue: I see not much of an issue here since this allows to have a similar rating change as with the Elo-System.

5. Quantity v
This describes the estimated variance of a team’s and/or player’s rating based only on outcomes of matches. It includes individual ratings and rating derivations.
It still gives me a bit confusion to what its purpose is.

6. Quantity “Delta”
This describes the estimated improvement in rating by the comparison of a pre-period rating to a performance rating based on the outcome of matches.
I assume this includes a rating before a period of no matches and a recent (and still active) period. This explains maybe why some players could take a break for a longer time and still not be punished by a much lower rating as there was a small decay.

Questions
How much of an issue is the convergance to a rating and can it be resolved in a way ?
Are 10-15 games enough to determine a rating and most of all an authentic one to display ?
Is this system a “carrot on a stick” as there need to be many games played to have a low Ratings derivation ? How does it make the leaderboards authentic and how does it not ?
How many factors play too impactful roles ? Balance ? Matchmaking ? Low decay ?
Does SoloQ need to have a leaderboard ? Can it be authentic ?

Evaluation: The systems has additional data to the ELO-System but falls off by the flaws the ELO-System also displays. Rating can be held high by a small decay.
There are also more factors which make it hard to pin down if a rating or ranking position difference between two decent players is authentic. Skill cannot easily be determined, especially by a numerical difference in rating.

Solutions:

*Make a Team-Queue in-game leaderboard only, keep Solo-Queue unrated, implement an unrated Team-Queue for casual players.
Make dailies and monthlies tied in to all queues (similar as it is now).

*Balance is still the most important issue and the reason things are so grim as the skill ceiling has been lowered with each patch by weapon and skill locking and band aid fixes which did not solve any crucial issues mentioned by decent players before.
Split it more, PvE, WvW and sPvP all suffer under mixed balance! Keep sPvP demanding of fast thinking and fast reactions. Make it difficult. Make it that there is a high skill ceiling and a high amount of complexity.

*A league system would do wonders. There is not competition about ranks anymore but there are rather groups of teams and players in a certain division. And rating would not be held by means of staying inactive. (Though a longer time of inactivity could mean a drop to a lower division)
With seasons, there can be rewards, e.g. a division-coloured star similar to the one for world-completion with maybe a number in the center of it displaying the season that the reward was given out.
And at the end of the season, highest division teams could be invited to a continental tournament to determine a few teams being able to go to an international tournament.

*No paradigms anymore about other groups of players being overwhelmed (what a bad excuse to keep more to learn out of the game). The more depth sPvP has and allows to learn something new, the more interesting it is to watch decent players dish out new builds, tricks, etc. and prolongs its life as a competitive game. It has to be said that the life of a competitive game is dependant on the competitive community.

Personal Evaluation: The leaderboards and its Glicko-2 system are nice on papers but fall off in practice compared to a league system two other games use.
Furthermore, other issues stated above give the leaderboards way less authenticity and kill off more competition.

the mere fact is that Arenanet does not care enough about pvp to implement their own sound ranking system like CS GO did that works. Nor did they even try. Our leaderboards is the biggest shortcut known to man its a JOKE!!!!

THIS GAME IS A JOKE!!!!! WHEN WILL THEY GET IT !!!!!!!!!!!!!! SUCCESS LIES IN SOLID RANKING AND MATCHMAKING SYSTEMS!!!!!!!!!

WHEN WILL THEY LEARN WHEN!!!!!!!!!!!!!!

They already said they are totally revamping the rating system into a tiered ladder like in Starcraft. Coming out sometime early this year. Patience, bros.

March update…. one can hope.

Aldizi.5671:

the mere fact is that Arenanet does not care enough about pvp to implement their own sound ranking system like CS GO did that works. Nor did they even try. Our leaderboards is the biggest shortcut known to man its a JOKE!!!!

THIS GAME IS A JOKE!!!!! WHEN WILL THEY GET IT !!!!!!!!!!!!!! SUCCESS LIES IN SOLID RANKING AND MATCHMAKING SYSTEMS!!!!!!!!!

WHEN WILL THEY LEARN WHEN!!!!!!!!!!!!!!

The matchmaking is perfectly fine. Any further tuning of it that could possibly have any chance of making it better would most likely lead to a fairly drastic increase in queue times. Some people would be willing to make that trade while others like myself would not. Really the only thing I think that would make matchmaking better is for the game to never look outside of a certain ratings range of players to fill out rosters regardless of time in queue. The rest is all open to debate as far as I am concerned maybe the ratings could be fine tuned a little more maybe not. The matchmaking/rating system is not what is killing pvp it is the small playerbase which is the is the direct result of lackluster balance, lack of game modes, and mediocre rewards imo. They have spent tons of time on matchmaking and there ratings system believe it or not.

@brannigan
I consider the 4vs5s (even if falsely) as a part of MMR and this is what makes me rather worry. I have luckily not experienced it on an other account before but I can very well understand the frustration and the need to fix it. It should be high priority after balance, game modes and leagues.