DUST Search - questions about the new matchmaking.

Ryme Intrinseca wrote:

emm kay wrote:

Ryme Intrinseca wrote:

emm kay wrote:

@ryme.
Ccp ratatti said himself that the algorithm was based closely on KDR.

CorrelationGëácausation

CCP Rattati wrote:

When enough players have queued up, a battle is created and the Team Balancer goes to work. Units are created out of individual players and squads, and these Units have the Rank of "sum of Mu of Unit", where Mu is a paramater closely correlated to both W/L and K/D ratios.

Once Ranking has taken place, Units are placed on Teams in descending order, always placing the Unit on the Team with the lowest current sum of Rank.

in this case, yes. It does.

No it doesn't. As Rattati says, it is correlated, i.e. there is a tendency for a high mu player to have high KDR as well. But mu is not based on KDR in any way. For that to be the case, KDR would have to feature in the mu equation. But the equation is this:

NewMu=OldMu+"Player seconds in Battle/Battle Duration" * Win/Loss

KDR doesn't come into it.


Author	Thread Statistics \| Show CCP posts - 0 post(s)
emm kay G0DS AM0NG MEN General Tso's Alliance 198	Posted - 2014.09.18 19:01:00 - [1] - Quote How about AFKers? Will they end up in a Mu stasis where they are matched up at the same str level because of the k/d not changing and the (almost) 50/50 win/loss? How about the logis? Logis have a low KDR. Will we be able to logi for a few rounds to be matched up with noobs? Or will we go logi vs. Logi battles? That brings up heavies They have the highest KDR and influence the WLR more than any other dropsuit type. Will we see the elites having huge bloodbaths withnno logi support? Speaking of bloodbaths, let's touch the sensitive topic of ads Everyone knows that the pilot doesn't get the kills, it's the gunner. Does this mean if you logi as a squad lead and get into lower battles that the pilot will get more oppertunities for the gunner? And finally squad items Will our 'battle rating' adjust slower if we are in squads? -- You called, sir?
emm kay G0DS AM0NG MEN General Tso's Alliance 199	Posted - 2014.09.19 02:23:00 - [2] - Quote @ryme. Ccp ratatti said himself that the algorithm was based closely on KDR. -- You called, sir?
emm kay G0DS AM0NG MEN General Tso's Alliance 199	Posted - 2014.09.19 11:04:00 - [3] - Quote Ryme Intrinseca wrote: emm kay wrote: @ryme. Ccp ratatti said himself that the algorithm was based closely on KDR. CorrelationGëácausation CCP Rattati wrote: When enough players have queued up, a battle is created and the Team Balancer goes to work. Units are created out of individual players and squads, and these Units have the Rank of "sum of Mu of Unit", where Mu is a paramater closely correlated to both W/L and K/D ratios. Once Ranking has taken place, Units are placed on Teams in descending order, always placing the Unit on the Team with the lowest current sum of Rank. in this case, yes. It does. -- You called, sir?
emm kay G0DS AM0NG MEN General Tso's Alliance 203	Posted - 2014.09.24 22:34:00 - [4] - Quote Ryme Intrinseca wrote: emm kay wrote: Ryme Intrinseca wrote: emm kay wrote: @ryme. Ccp ratatti said himself that the algorithm was based closely on KDR. CorrelationGëácausation CCP Rattati wrote: When enough players have queued up, a battle is created and the Team Balancer goes to work. Units are created out of individual players and squads, and these Units have the Rank of "sum of Mu of Unit", where Mu is a paramater closely correlated to both W/L and K/D ratios. Once Ranking has taken place, Units are placed on Teams in descending order, always placing the Unit on the Team with the lowest current sum of Rank. in this case, yes. It does. No it doesn't. As Rattati says, it is correlated, i.e. there is a tendency for a high mu player to have high KDR as well. But mu is not based on KDR in any way. For that to be the case, KDR would have to feature in the mu equation. But the equation is this: NewMu=OldMu+"Player seconds in Battle/Battle Duration" * Win/Loss KDR doesn't come into it. So, lets take a step back. I say that the algorithm is based closely on kdr. You repeal with the fact that correlation is not always causation I post a quote by ratatti, and say that in this case, it is. You form a rebuttal by telling me how close the correlation is with kdr? -- You called, sir?