Can someone who knows WvW math better explain to me what is going on here? Some of the changes make sense, like Fort Aspenwood trading spots with Maguuma (if you thought I hated Season One, talk to my buddy, the casual player from Maguuma. “I zone in and get run over by like 30 people”), and you see other servers trading spots, a few trading tiers in the process. And then Dragonbrand is being put in tier 5.
Hey, that’s our tier. Was it something I said? This is like week 8 of Season One, because DB is bringing as many WvW players as NSP and IOJ combined. The discussion thread includes DB WvWers reacting to their swelled numbers from casual folks soaking up points in the “free win” week. IOJ, remember, is one of the servers trying to fall to its new level after most of its WvW guilds transferred out during for Season One. Well, this will certainly help, except that NSP seems to be even more depressed than us.
I just don’t see any number on that chart by which DB fits in the same tier as IOJ and NSP. They even have a little green arrow saying they are heading up. So they drop two tiers but not change in ranking? I don’t claim to know the math, but it does not seem to be working, unless there is just a bit of randomization to “spice things up.”
: Zubon
“…unless there is just a bit of randomization to “spice things up.—
That is precisely what is happening. Each server gets a RNG modifier (based on some arcane hoodoo) added to their rating score, the match ups are determined by this adjusted rating. In this case it would seem that DB “rolled” low, while IoJ and NS “rolled” high, thus putting their adjusted ratings in the same bracket.
This was done several months ago because the match ups were getting rather stagnant, and it was thought that by shaking things up a bit, things would be more interesting by allowing servers to fight something other than the same two servers they had been fighting for months and allow for tactics and strategies to propigate through the tiers. At the same time the ratings could become more accurate through a larger sample size of match ups for the statistical algorithms to munch upon.
Arenanet just made an official post about this system: https://www.guildwars2.com/en/news/wvw-matchup-variance-reduction/
Curiously, it’s to announce that, starting with this matchup, the possible variance was significantly reduced.
Well firstly there’s the whole modified Glicko thing that I totally failed to understand from the get-go. For as long as WvW has been running, under any of the variations in scoring, our server’s actual performance during any given match has borne little logical connection with the green or red up or down arrows beside it on Mos millennium. We often came third in our match only to see our rating go up, or won only to see it drop.
Then, as Pi explains, there was a change a while back that added a random element, which from memory allows servers to be drawn either a tier above or a tier below their actual tier, giving each match a potential pool of nine servers except for the top and bottom tier. That was the standard for a good long while before the Season so unless you weren’t playing then you must already have experienced it – maybe IoJ just had a series of unexceptional random draws?
From a Yak’s Bend perspective there’s a very simple problem. We are, and have for a very long time been, *exactly* right for our slot. We are predictably either the weakest team in T3 or the strongest team in T4. Because we have an extremely stable WvW population – we still have mostly the same guilds, Commanders and regular players turning out that we had a year ago – we feel like the rock that the river flows around. Nothing much ever changes on YB and we like it that way.
It does mean, however, that we are caught in an endless cycle of rather meaningless matches. If we do well we go up, then we get pounded and drop. Then we pound who we fell on an go back up. Or sometimes we just stay put for week after week after week with the teams around us changing not all that much. I can’t see the point.
That above all is why I loved the Season so much. It added a much-needed structure and a sense of completion that all the other variations have lacked.
(Although why anyone should listen to anything I say when I can’t even spell my own name…)
The rating system is essentially like spread betting in sports. The algorithm determines what the expected outcome of a match up should be based on past performance, if you do better than expected your rating goes up, you beat the spread and won, do worse than expected and lose rating.