I’m not saying they’re using it exactly as it is.
The same way TBC wasn’t using the Elo system exactly as it was.
But they basically copied most of it. As is. Then they threw in teams instead, in TBC.
Uhm… No.
That has to do with the hidden personal MMR of each person.
And how their own MMR can be destabilized or not, and so on.
Okay, firstly - whether it’s ELO or Elo doesn’t matter, and a lot of people do actually write it “ELO”. So much so that it’s acknowledged on Wikipedia.
They took inspiration, sure, but it is not the same system, and what that means is that you cannot just take facts from it a la carte and say “Because it is this way here, therefore it works like that in WoW”.
That’s exactly right. When you make a new team it uses the hidden MMR’s of every player to try to work out where to place you, and it’s very lenient at first, trying an almost random guess anywhere in the range of the players in it. Often, however, it’ll try out around 1500 first even if all 3 players have 2.4k MMR. Like it’ll write you have 2.3k MMR and then it’ll still match you against a 1.6k team or something.
No, that phenomenon has to do with the deviation value being out of whack for 1 or more of them.
Queue up with new people normally when you have 2k yourself, with them having a similar win rate as you (not from a losing streak), and CR, and you’ll see it ending up as if you never swapped out the people you last played with.
Swap out that 2k with whatever you want, and it’ll still be the same effect as described.
It has to do with its inherent volatility.
That’s just not what happens. For several expansions I played with 3 different priests all between 2100 and 2200 and every time I swapped one for another for the first time in a season, even though they were all at 2100, we got to restart. Once I played regularly with them all though, the rating stabilised for them all.
You don’t know the system they’re using. Stop pretending you do. And I’ve got to get back to work.
I’ve done that kind of stuff myself, and the things I’ve seen confirms what I know of the glicko system.
I don’t think you’ll find many people saying they’ve met 1.6k players when they’ve swapped people out at 2.1-2.2k with themselves having the same.
Wtf m8 are we even playing same game? Why is my experience different from yours?
Ppl in this forum must be trolling or idk.
Cuz he’s talking about a weird result of people’s volatility being off. It’s not common. Can still happen, but is really not common.
Unless he was talking about shifts from season to season, then there’s a weird remnant of the previous season if you get past a certain threshold and so on, or if 1 or more of the other people he was teaming up with were far below his own rating then it can also average out that far below.
Anyway, either way, personal MMR is king even though you can’t see it, and it has its own deviation value attached to it which determines its volatility which in turn controls how “far” it can jump from game to game and so on, and what you win/lose against also affects the deviation value, as well as win streaks/losing streaks etc.
People confuse the ELO as being short for something, which it isn’t. It’s named after its creator, thus you should write it as Elo. The same way Glicko is named after its creator, Glickman.
But to summarize, ofc I know they’ve made changes of their own to the system. But the bulk of it is still identical to the Glicko system.
And there’s no lenient period of 5 games whenever you team up with new people. The placement games when starting from 0 are forcibly made much more volatile, but when you just queue up with new people after already having cleared the placement games, then it’s just your own personal MMR and its own volatility being matched up with the other people’s personal MMRs and their own current volatility, and then it smooths out the differences slowly over the course of several games but it doesn’t start with being more lenient. It keeps going for the median, since the median number is representative of your combined abilities up to that point, which is then further affected by the games that you keep playing, with each person losing/winning different numbers if you all started on different ratings, to then end up with the same after a lot of games (not just 5), and the streaks of winning/losing makes that “smoothing out” happen much faster. But there’s no forced volatility like in the placement games, it’s just a natural consequence of personal MMR and personal volatility meshing with other people who start from different places.
You might just not be noticing it, idk. It happens quite a lot.
When you find a new partner it takes all of one unexpected loss to send the volatility off the charts.
At low rating you don’t generally experience meeting higher rated players that often however, but that’s simply because there’s far less of them and it’s a very small subset of their games (because most of them aren’t PuGging either) where they get this volatility.
RBG’s have a famously large amount of this kind of volatility because the comps swap constantly due to team size. It’s not uncommon for MMR in RBG’s to jump hundreds up and down with every game. I’ve tried being 1.8k and then 6 games later been 2.6k and then 4 games later been 2.2k.
I’m not. If you play with the same buddies a lot though you won’t really see it though. I’m not suggesting that it happens every or even most sessions. It only happens when you start playing with a new team and have an early unexpected loss or win.
Maybe you’ll get a little volatility on new seasons (it increases it when the seasons swap and it also lowers your MMR equal to the inflation you got from last season so it’s back to normal)
It is 5 games until it settles down and starts spending a lot of time before it widens the matchmaking.
You can see this in StarCraft 2 as well, where the implementation of the system is a little more up-front and in-your-face. You can see your personal MMR and you can see your placement games being marked as placement games. It’s the same system except for the aforementioned inflation.
Yes, I know. But that has more to do with the system is trying to still calibrate the team’s combined ability, and it’s like you’re throwing a wrench into the wheel when you deviate from the set pattern you’ve been showing it thus far.
It’s not like the placement games when you start from 0. It’s not that the first 5 games are forcibly volatile like that, as you describe it.
The Glicko system explains it better, which would be a huge wall of text if I were to explain it now and you know you’re capable as much as I know you’re capable of just looking that up yourself.
(It’s better if you keep imagining your own hidden personal MMR and the one(s) you’re playing with and their personal MMR, and how that would look like after each game as you keep playing together, then it makes more sense with how the rating fluctuates the way that it does.)
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.