There is something wrong here.

[Jellby]

... that could literally apply to anything (except, I guess, something with which nothing is wrong).

Like... the RNG?

[Romain672]

But instead of criticising the subject, we can try to find ideas on how to resolve the problem.
Actually the system is really bad for team elo, but if we propose a system slightly better, that will not be good because changing the way elo work can make others problems. If instead we propose a good system, that should be good.

So the problems are:
1) like I said in my previous post, with AB vs CD, as A, the elo of B should matter
2) if B abandon, you need to do something with the other elo. If you gave a full victory, that could cause problems with the top 20 of the game, especially on high lucked games
3) with ABCDE vs F, if F lose, A is supposed to gain elo, even if it's low. It's not the case actually, there is many cases where actually A gain negative elo because he tied with BCDE (which is changed to 0)
4) if ABCDE tied for first place and F is second, it seem it should work differently than the previous point
5) if possible the system should not be too complicated to be able to be explained

And that's all that came to my mind.
So if someone got a solution which resolve all of those, that would encourage an admin to look into it.

Points 2), 4), and 5) work actually.
Point 1) and 3) does not work well.
I think in a team game, you want to calculate for each player a certain %. The elo of the teammates of each player will modify that %. At the end of the game, the elo gained or lost will ignore the teammates, but will be multiplied by that %.

[ChiefPointThief]

For instance, here is just one game of bang (There are several. It actually occurs a lot which is sad. Also people playing from multiple accounts in the same game). The player who was already eliminated (renegade. lone player. impossible to win) got elo. The player who was a teammate of the rage quitter got elo. Even my teammate got 17 elo. I got nothing. It isn't even like I have hundreds more elo than these players. We are all within 100pts of eachother.

https://boardgamearena.com/7/bang?table=394769850

[Jellby]

You got +0.73

[ChiefPointThief]

You got +0.73

Anything under 1 shows as 0. Maybe I've been getting like .1 for some of these games I thought I was getting nothing. The quitters made me remember this topic. After I posted here I opened another game which was rage quit again. I go to the account and like a lot of other times in this game it is a new account. I put this in suggestions but I would really like the option to not have new accounts at my table. I already do average and above but anyone with 2 wins qualifies for that category. The thumb system doesn't work. I've red thumbed several bang cheaters recently and they just keep coming. Some of those even deserve public red thumbs but then you have to find 5 recent players to green thumb. Bad system

[Ceaseless]

For instance, here is just one game of bang (There are several. It actually occurs a lot which is sad. Also people playing from multiple accounts in the same game). The player who was already eliminated (renegade. lone player. impossible to win) got elo. The player who was a teammate of the rage quitter got elo. Even my teammate got 17 elo. I got nothing. It isn't even like I have hundreds more elo than these players. We are all within 100pts of eachother.

https://boardgamearena.com/7/bang?table=394769850

The difference between you and most of these players was 3-4 points. The 17 came from a newer account with higher variation.

Though if you run into boosters you should report those, I've had to clear a group of those before in Bang, when you have multiple accounts working together (often across various sides) so a specific account goes up. It becomes very obvious when the big account is eliminated and cannot win and so one of the other boost accounts times out so that big account "wins" the game.

[Romain672]

Chief gained 2.58 elo from the people which leave.
But he tied for first place against the four others players:
- he lost 0.29 against d_ario
- lost 0.58 against sovenok
- lost 0.99 against cat228

2.58-0.29-0.58-0.99 = 0.72

We was in my point 2) from before:

2) if B abandon, you need to do something with the other elo. If you gave a full victory, that could cause problems with the top 20 of the game, especially on high lucked games

With a second though, if Chief gained 2.58 elo (ie he win the game against the leaver) and others elos were ignored, that could work nicely.
edit: but k-factor of that game was 7. It's weird it stay at 7. If you put it back to 20 (like in 1v1 and 1v1v1 i think) you would gain 3x more elo.
In a good part of the leaves, the person which left was losing. So it look good to assume he lost against everyone else.