Loading...
  OR  Zero-K Name:    Password:   

FFA games and rating

14 posts, 735 views
Post comment
Filter:    Player:  
sort
All free-for-all games played before yesterday which were tagged for Casual rating have been untagged for that rating and moved into their own category. Since there was also a server restart (with an update to expose mapoptions as well as modoptions to lobby if I understand correctly, thanks CZrankAdminLicho) this change has been applied to everybody's "current" casual rating, although your "ladder" rating may take a while to catch up.

---

Why was this done?
There have been concerns that some people were using FFA games to (a) dump their own rating to get easier team games, and (b) throwing their rating to friends.

It's difficult to tell from comparing previous and new ratings whether this manipulation was particularly successful.

Even if it was not successful, the potential for people to attempt it (even the idea that other people might try to attempt it) did not seem to be healthy for the FFA community.

Previous analysis suggests that this change has probably, on average, decreased the accuracy of Casual ratings a bit. At this stage that seems like a reasonable price to pay for fixing this FFA community issue.

---

How can I look at the new category of FFA games and get a rating etc?
At the moment, you can't (although Dave's third party system should still work). I don't think it would be a bad idea to have some kind of way to track FFA performance, but it doesn't exist right now. A full WHR ladder sounds like overkill (and I'm not sure what impact it would have on server performance). Maybe calculating stats once a month is a reasonable idea. If the API that's been talked about recently comes to fruition then it should be pretty trivial to look at this category of games.

---

Why do a lot of players now have hundreds more casual rating?
I'm not really sure. WHR is a complicated system. Maybe DeinFreund could guess.

---

What will happen to new FFA games?
At the moment new FFA games may go towards casual rating. You can expect such games to be put in the same bucket as the existing FFA games at some point in the future, though.

---

What will happen to no-elo games on the FFA autohost?
I believe we have the capability to move these into the new category also. At this stage I don't have a strong opinion on whether we should do that.
+1 / -0

20 months ago
All of this so far has been about the effect which FFA games have on rating. A separate question is how rating should affect FFA games. Putting aside the effects of "let's all gang up on the purple" diplomacy, the actual numerical ratings do not affect how a "classic" FFA game is configured by the server and game.

However, ratings obviously do affect how teams are set up in a teams-FFA context. My personal opinion is that even if a full FFA WHR ladder were set up, we'd still probably be better off balancing teams-FFA according (mostly or entirely) to everybody's casual rating, since the FFA rating data is likely to be very poor quality. Maybe it's possible to set up a Casual+FFA ladder to balance FFA, while keeping a pure Casual ladder to balance normal teams. Not sure.

I have very little to do with teams-FFA so this is a topic on which I think discussion from those with experience might do some good.
+0 / -0


20 months ago
Wow, I did NOT expect to discover that the reason casual ladder has lower elo than MM is FFA. :o

This could do awful things to lobsterpot balance for those who've gained 400 elo like me...
+0 / -0
quote:
Wow, I did NOT expect to discover that the reason casual ladder has lower elo than MM is FFA. :o

The top of the currently active ladder still has considerably higher MM ratings... but the currently inactive Godde has current casual rating in the 3600s.

I'm not sure what higher or lower rating points compared between two separate WHR systems actually means, though. It may mean something, it may not. I guess I could imagine that FFA is more "random noise" than even teams, which caused the system to be less confident in predictions across the board.

quote:
This could do awful things to lobsterpot balance for those who've gained 400 elo like me...

Four hundred is on the high side I think, but gaining several hundred rating points seems to be the rule for most players rather than the exception. So I expect the impact to not be extreme.
+1 / -0

20 months ago
I just removed about 150 games played in the "Casual 1v1 (no elo)" host from casual rating, where they had been erroneously placed due to server resets messing with autohost flags. This probably won't affect anybody's rating until the next server reset.
+0 / -0


20 months ago
What's the official word on removing 1v1s from Casual entirely?
+2 / -0


19 months ago
Bump for sustained interest in an answer to the above question.

Also, worth noting that FFA has been much, much less popular since this change. Probably correlated to how much people enjoy gambling their rating. If correct, another reason for the 3 ladder system.
+4 / -0
quote:
Also, worth noting that FFA has been much, much less popular since this change. Probably correlated to how much people enjoy gambling their rating. If correct, another reason for the 3 ladder system.

FFA has fluctuated from domination to desert in the past with no changes to how ratings worked, so i don't think this is a safe inference to make.

One or two regulars no longer seeding, and it's off.
+2 / -0
19 months ago
Except for high ranked 1v1 players, I can say that player rank now feels more accurate in teams room. Much less purple players with FFA inflated rating.
+0 / -0


19 months ago
Didn't removing FFA make people's ratings go UP?
Methinks your statement is groundless.
+0 / -0
19 months ago
I just wanted to subtly bump the topic, it took me an embarrassingly long time to come up with that.
+0 / -0
Removing FFA games caused a general recalibration, so most people going up in rating doesn't mean much. The important part is whose rating did something unusual.

On that point, I don't think many people's rating went down tremendously (which would have implied that their rating was unreasonably inflated by FFA games). Most of the purple players who I would have categorised as FFA grinders are still purple, or close to it.

However, a few people's rating went UP a lot (suggesting that they were actually losing a lot of FFA games, keeping their casual rating down).
+2 / -0


19 months ago
Agree with the above. I think just about the only "victim" in this was unknownrankChomolungma.
+0 / -0
19 months ago
#ChomoOnTop
+2 / -0