Fix your ELO

[Den_Man]

[/quote]
I can mostly agree that ELO is meaningless and still question why someone would care about that metric if they hold that belief.
[/quote]

I think the point of this thread is that ELO is meaningless in the way it is implemented.

[mturton]

I wish I could simply turn off ELO. The system can keep it to match me with people, I don't have a problem. But I hate seeing that number there, it affects whether I want to play the game ("Oh no, my ELO might fall").

Can you guys give us an option to turn it off?

[Meeplelowda]

I can mostly agree that ELO is meaningless and still question why someone would care about that metric if they hold that belief.

I think the point of this thread is that ELO is meaningless in the way it is implemented.

We obviously aren't talking about ELO in the abstract or ELO as used specifically in chess federations, so "in the way it is implemented [on BGA]" is implied. So if ELO in the way it is implemented on BGA is meaningless, then why would one alter one's behavior in light of it?

This is separate from the question of what could be done to change ELO in the way it is implemented on BGA (see how cumbersome that is to say it every time) so that it is no longer meaningless. There are other threads about how to tweak ELO in the way it is implemented on BGA and I've expressed support for some of the suggestions (for example, Silene's suggestion on the first page of this thread).

[HRG_Surplus]

So he is a prime example of what I am trying to explain. I have an ELO of 100 and I was 3 points away from a player of 238 ELO by 3 points. But I get deducted by -30 points negative. But Doggie lost from a 198 ELO and only went down -4 points. However Barton has a 198 ELO that beats a 238 but only receives a +15. This whole thing makes no sense. I should lose at basis less than 10 points for coming 3 points shy. I was 1st player as well. And that should had offset the point age by a lot. If I was last player I would had beaten doggie by 1 point at least. Fix YOUR SYSTEM ASAP!!!! Please or please explain this to me 1 on 1 so I can question and see how it makes sense. Geez.

[Meeplelowda]

So he is a prime example of what I am trying to explain. I have an ELO of 100 and I was 3 points away from a player of 238 ELO by 3 points. But I get deducted by -30 points negative. But Doggie lost from a 198 ELO and only went down -4 points. However Barton has a 198 ELO that beats a 238 but only receives a +15. This whole thing makes no sense. I should lose at basis less than 10 points for coming 3 points shy. I was 1st player as well. And that should had offset the point age by a lot. If I was last player I would had beaten doggie by 1 point at least. Fix YOUR SYSTEM ASAP!!!! Please or please explain this to me 1 on 1 so I can question and see how it makes sense. Geez.

1) What was your K factor for that game? If you don't know what that is, there are FAQs explaining how ELO is calculated.
2) Losing by only 3 points away doesn't matter. No ELO system factors in how "close" your loss was. Win/lose/tie is all that matters.
3) You lost to two people. Both losses factor in your adjustment.

[HRG_Surplus]

So here is the final example, how does this make sense?
Game: It's a Wonderful World
K Factor 15
I have a ELO of 134
Person A has ELO 234
Person B has ELO 205

Person A came in 1st place
Myself came in 2nd Place
Person B came in 3rd Place

I only made +4 on that, how is that even fair???

This is absolutely out of this world. But if I go against lower level players my K factor is higher. That seems to be the logic here. If I am playing lower levels I get a higher K Factor. But there are more higher level players though. So the chances on getting lower level ELO players are smaller in ratio than to higher ELO players. It's hard if not impossible to reach to the top or even have a balance of playing players at my own level which is the issue here. That is including Tournament play. If I am stacked against it's entirely impossible to win unless some outside factors in cyberspace occurs to deter the outcome. It just looks like that since I was keeping a keen eye on ELO ratings and K factors including how many players are playing in the game.

[Silene]

So here is the final example, how does this make sense?
Game: It's a Wonderful World
K Factor 15
I have a ELO of 134
Person A has ELO 234
Person B has ELO 205

Person A came in 1st place
Myself came in 2nd Place
Person B came in 3rd Place

I only made +4 on that, how is that even fair???

This is absolutely out of this world.

So you are 2nd in a 3-player game. That's a very average result. If your opponnents were the same level as you, it would result in +-0 ELO. But since your opponnents are a little higher ranked than you (~ +85), you have a little ELO-gain from this game.

So what is not fair? Or "out of this world"? Seems pretty fair to me...

(Please don't compare this with your previous example which was for Gizmos where you still had less than 10 games and therefore had a high K-factor which is not the case for this game here. Your K-factor does not depend on who you play against. It only depends on the game being played and the amount of games you have already played in it)

[Dr Tumbleshlong]

I find it almost impossible to advance in ELO-ranking. I'm 75 games in 6NIMMT, and I went past ELO 300 after 13 games. Due to a fairly good start my average shows +18,6 as average score. I don't get it. My average scores have been:

First 13 games I averaged +24,2 points per game.
Following 62 games I averaged -0,27 points per game.
In my 75 games I have averaged 3,97 points per game.

I mainly play 5 player games (few 4 and 6 player games included)

In my last 15 games (averaging +3,3 points) I have been:
1st: 6 times
2nd: 2 times
3rd: 3 times
4th: 3 times
5th: 1 time (6 player game)
Last place: 0 times (regardless of amount of players)

Placing 1st: +19,8 points avg
Placing 2nd: +2,5 points avg
Placing 3rd: -2,0 points avg
Placing second last: -17 points avg
Placing last: no idea

Get the picture? I can't even lose a game, if I were trying to rank up. Placing in the TOP2 nearly 55% of the time is barely positive in result.

It seems that the weight for the first 10 games or so is flawed and the emphasis is way off. And I'm not even trying to understand why there is an emphasis if it is already broken by design.

It should be at least a more decent spread, for example:
1st place avg +20 points
2nd place avg +7 points
3rd place avg anywhere around +/- 5 points
4th place avg -7 points
5th place avg -20 points

This was the latest (screenshot) where after a tight game last round cards set the tone and and three players - all within the same hundred mark in ELO ranks - score 51 points spread in ELO points, even we all three nearly tied the score. Just ludicrous.

And please don't give me that K factor -rubbish. That's part of the problem. The ELO ranking system (atleast in 6NIMMT and in Carcassonne as well) needs serious remodeling.


Apparently I don't know how to add a picture here, so here's the link to the screenshot: https://imgur.com/a/laqixPS

[Tisaac]

I find it almost impossible to advance in ELO-ranking. I'm 75 games in 6NIMMT, and I went past ELO 300 after 13 games. Due to a fairly good start my average shows +18,6 as average score. I don't get it. My average scores have been:

First 13 games I averaged +24,2 points per game.
Following 62 games I averaged -0,27 points per game.
In my 75 games I have averaged 3,97 points per game.

I mainly play 5 player games (few 4 and 6 player games included)

In my last 15 games (averaging +3,3 points) I have been:
1st: 6 times
2nd: 2 times
3rd: 3 times
4th: 3 times
5th: 1 time (6 player game)
Last place: 0 times (regardless of amount of players)

Placing 1st: +19,8 points avg
Placing 2nd: +2,5 points avg
Placing 3rd: -2,0 points avg
Placing second last: -17 points avg
Placing last: no idea

Get the picture? I can't even lose a game, if I were trying to rank up. Placing in the TOP2 nearly 55% of the time is barely positive in result.

It seems that the weight for the first 10 games or so is flawed and the emphasis is way off. And I'm not even trying to understand why there is an emphasis if it is already broken by design.

It should be at least a more decent spread, for example:
1st place avg +20 points
2nd place avg +7 points
3rd place avg anywhere around +/- 5 points
4th place avg -7 points
5th place avg -20 points

This was the latest (screenshot) where after a tight game last round cards set the tone and and three players - all within the same hundred mark in ELO ranks - score 51 points spread in ELO points, even we all three nearly tied the score. Just ludicrous.

And please don't give me that K factor -rubbish. That's part of the problem. The ELO ranking system (atleast in 6NIMMT and in Carcassonne as well) needs serious remodeling.


Apparently I don't know how to add a picture here, so here's the link to the screenshot: https://imgur.com/a/laqixPS

I dont get it. You are not talking at all about the opponent average ELO in your analysis where it's the only thing that maters. If you have 1st place against only small ELO, then yeah it makes sense to have such elo adjustments

[Romain672]

You could have alternative system on high luck games, per example by making the difference of level matter less.
But to reach high elo you would need more games to play, and it makes one extra step to explain elo to someone which isn't familiar with it. That's two downsides already.