Jump to content
  • Sign Up

http://gw2stats.com/na <--- ANET for you! *READ*


Kal Durak.2830

Recommended Posts

12 minutes ago, Kayowin.9217 said:

And only based on the 8th weeks performance.   More Anet quality content coming to a BL near you!

Glicko only needs 4 weeks to "settle" at minimum with such small number of servers.  This was per a dev post when relinks were introduced and the length of time was set to 8 weeks before relinking to give time for that settling.

The servers that are linked and become host have outdated ratings, like Sea of Sorrows, which throws the system into a loop.  But regardless of that, if you look at the ratings spread even with the outdated ratings from former links, the difference between the top rated and lowest rated server is less than 100, well within the +/-100 randomization roll.  That is nowhere near how it used to be when glicko was used for all matchmaking and servers had spreads way larger than 100.  That speaks volumes to how the teams  have evened out over the years.

It's an illusion only in the mind that there's large imbalances even with this being the first match and random.  Check out the current Victory Points scores right now across all NA matches.  Two out of the four tiers are quite close and there's still enough time in the match for the lower score to catch up if they wanted.  The larger imbalances lean more towards skill and coverage issues than playhour totals.

Edited by Chaba.5410
  • Thanks 3
  • Confused 1
Link to comment
Share on other sites

Thanks Chaba.   

I just find it amazing how a certain server can tank week 8 in T1 then drop all the way to T4 for week 1 of the new linkups.   Yes they don't have the strong link anymore propped up by a certain commander, but I still call BS on Anet's part.   

  • Like 2
Link to comment
Share on other sites

3 hours ago, Chaba.5410 said:

What are you trying to say? It's a glicko match. The first match after relinks is always random and wild.

When looking at it from a coding angle, the weeks of chaos after relinks makes players question the logic that will be used in the WR project after new server/shards are generated and how well the coding will stand up. Considering host servers positions before and after match links if the secondary servers play that much of a factor in positioning it makes people question why the links? Using glicko, shouldn't the hosts stayed in the tiers they were in? Considering the weekend has more activity then the weekdays the spreads are pretty broad already , T1 17, T2 50, T3 42, T4 24. When looking at the placements at 1 shows not just coverage but population and those are even more distinct. No, I question their placement logic here after sorts myself. Each to their own but doesn't leave me with much comfort in the logic.

Link to comment
Share on other sites

19 minutes ago, Xenesis.6389 said:

Not when a medium link server has the highest rating...

That's because the rating hasn't been updated since 4/1/23, which I mentioned in my earlier post.  You need to look at only the host servers in that list.  It's still far more informative than the gw2stats site which is a big mess.

Edited by Chaba.5410
  • Confused 1
Link to comment
Share on other sites

19 minutes ago, TheGrimm.5624 said:

Using glicko, shouldn't the hosts stayed in the tiers they were in?

Recall that glicko matchmaking had a +/-100 randomizer thrown into it.  So no.  The matchmaker takes the rating, rolls a random number between +100 and -100 and adjusts the rating with that, then sorts the servers by the randomized rating.  There used to be websites that calculated the probabilities of which rank a server would end up as, the probability of which tier and color in the tier and against who.

  • Confused 1
Link to comment
Share on other sites

5 minutes ago, Chaba.5410 said:

Recall that glicko matchmaking had a +/-100 randomizer thrown into it.  So no.  The matchmaker takes the rating, rolls a random number between +100 and -100 and adjusts the rating with that, then sorts the servers by the randomized rating.  There used to be websites that calculated the probabilities of which rank a server would end up as, the probability of which tier and color in the tier and against who.

Fair, but you only make it seem worse in hopes of the new system. It's a holiday versus a work day and I have some work to do in any case 🙂  else you would make me grab a spread sheet to try and back myself into their coding logic and the different servers glickos to try and see how much random applied and if that shouldn't be removed. Looking at the point spreads its going to be some lopsided matches as servers resort.

  • Like 1
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...