Another complaint about ELO

[frogstar_A]

Actually, your representation go the 4 player game does not reflect what I would like to see. Ideally, in a four person game of equal ELO players, where 10 pts is the basis, the winner would get 30 and everyone else would lose 10. Because I don't think it should matter if you're 2nd out of 4. 2nd in a tournament? Sure, you should get a medal for that. 2nd in arena standings? Again, well done. Hell, top 10 in that is great. But out of 4? In a single game? Not so much.

What's worse would be that I'm guessing the reason why I only got 12 points for winning the game in question was because the "wrong" person got 2nd. Is it safe to assume that, had the highest rated player gotten 2nd and the two lower rated players finished 3rd and 4th, those 2 players wouldn't have bled as many ELO points and I would have gotten more than 12? Which invites all sorts of ugliness. Either a) this is something that I can't affect so should not alter my score or b) I ethically shouldn't try to affect. Because surely you don't want players trying to affect the chances of the rest of the competitors finishing 2nd or 3rd or 4th. Do you?

And, if it is your intention to have ELO reflect the skill level of players, then why do you penalize players for playing multi opponent games as compared to H2H. And the math supports that. If you need me to show you, I will.

YOU might not think 2nd is any better than 4th, but a good ranking system should surely reward a player whom is consistently 2nd of 4 higher than a player who is consistently 4th of 4?

Plus under your system you might as well give up the moment you can't win, instead of having 2nd to aim for. Its a worse system that would create worse play experience for most people.

[ahyangyi]

The basic assumption of Elo on BGA is that players play for rank, not winning.
Occasionally this isn't the designer's intention, and they will implement a winner-takes-all style Elo.

But not really for a normal game like Wingspan.

[detlefchef11]

Actually, your representation go the 4 player game does not reflect what I would like to see. Ideally, in a four person game of equal ELO players, where 10 pts is the basis, the winner would get 30 and everyone else would lose 10. Because I don't think it should matter if you're 2nd out of 4. 2nd in a tournament? Sure, you should get a medal for that. 2nd in arena standings? Again, well done. Hell, top 10 in that is great. But out of 4? In a single game? Not so much.

What's worse would be that I'm guessing the reason why I only got 12 points for winning the game in question was because the "wrong" person got 2nd. Is it safe to assume that, had the highest rated player gotten 2nd and the two lower rated players finished 3rd and 4th, those 2 players wouldn't have bled as many ELO points and I would have gotten more than 12? Which invites all sorts of ugliness. Either a) this is something that I can't affect so should not alter my score or b) I ethically shouldn't try to affect. Because surely you don't want players trying to affect the chances of the rest of the competitors finishing 2nd or 3rd or 4th. Do you?

And, if it is your intention to have ELO reflect the skill level of players, then why do you penalize players for playing multi opponent games as compared to H2H. And the math supports that. If you need me to show you, I will.

YOU might not think 2nd is any better than 4th, but a good ranking system should surely reward a player whom is consistently 2nd of 4 higher than a player who is consistently 4th of 4?

Plus under your system you might as well give up the moment you can't win, instead of having 2nd to aim for. Its a worse system that would create worse play experience for most people.

Well, my compromise of winner getting points from 3rd and 4th, and 2nd place not losing or gaining any points satisfies that. And assures that the person who actually wins the game walks away with the most points.

And, before you ask. No, I don't think it's worth designing a system that differentiates between 3rd, 4th, and 5th in a 5 player game. Sure, throw 2nd a bone to make it worth it for people who need a consolation prize to keep trying. But 3rd? c'mon!

[detlefchef11]

The basic assumption of Elo on BGA is that players play for rank, not winning.

That's an unfortunate assumption.

As I've stated above several times, I can get with a consolation prize for second. But I'd imagine that the vast majority play to win.

[frogstar_A]

And, before you ask. No, I don't think it's worth designing a system that differentiates between 3rd, 4th, and 5th in a 5 player game. Sure, throw 2nd a bone to make it worth it for people who need a consolation prize to keep trying. But 3rd? c'mon!

In a 7 player game of 7 wonders coming 3rd is pretty good and coming 7th is rubbish. Those things should be taken into account in any good rating system.

All the reasons why it matters to differntiate 1st and 2nd place ALSO apply to every other place.

ELO is not a perfect system, but any system that ignores placings below 1st or 2nd is just worse. Sure everyone PLAYS to win but the rating system shouldn't Care what you were TRYING to do, only what you did relative to others whether that is better or worse.

If you only personally only care about winning then maybe just ignore ELO and look at your win % or number of wins....

[Silene]

Winner takes all is possible to implement within the game by setting all nonwinners on equal score during endgame scoring. So in games where that is not done, elo shouldn't override the differentiation in scores.

Lewis and Clark is doing winner takes all i.e.

Unfortunately it's done a bit wrong by considering the loser as tied. They should be considered as team losers. Or was that fixed with the update?

So whenever author/dev considers it to lead to a better game, it can be used. Otherwise elo should consider placement. Only doing it for 2nd seems artificial and unnecessarily complicated.

[detlefchef11]

And, before you ask. No, I don't think it's worth designing a system that differentiates between 3rd, 4th, and 5th in a 5 player game. Sure, throw 2nd a bone to make it worth it for people who need a consolation prize to keep trying. But 3rd? c'mon!

In a 7 player game of 7 wonders coming 3rd is pretty good and coming 7th is rubbish. Those things should be taken into account in any good rating system.

All the reasons why it matters to differntiate 1st and 2nd place ALSO apply to every other place.

ELO is not a perfect system, but any system that ignores placings below 1st or 2nd is just worse. Sure everyone PLAYS to win but the rating system shouldn't Care what you were TRYING to do, only what you did relative to others whether that is better or worse.

If you only personally only care about winning then maybe just ignore ELO and look at your win % or number of wins....

First off, if I drop this, it's because I've already gone through all this with others and don't want to just have the same debate with someone else.

As I've said several times already. While I prefer winner take all, I do understand that people want to spread the wealth among the other places so it's not the same if you finish 2nd as 4th, or I guess 3rd or 7th. But I absolutely feel the current system goes way overboard giving runner up points at the expense of the person who actually won the game. That is really it.

As I said in my first post, I won a 4 player game that included a player with nearly 200 more ELO pts than I did and got SIGNIFICANTLY less ELO pts than I would have had I beaten that player head to head. Someone then felt compelled to explain that, had that player (and I) had even higher ELOs and played two players who had never played before, it MIGHT have actually been easier for me to beat that high ELO player in the 4 player game than I would have H2H. Basically, they went straight to an entirely unrealistic scenario rather than actually engage me and discuss a game that had actually just happened.

So, sure, I suppose it's really neat to finish 3rd in a 7 player game. I mean, at least you're in the top half. That's what we call "a moral victory". However, in the game I posted about, someone got a pt for finishing 3rd out of 4th (and they didn't even finish above their seed in the game).

I guess I'm realizing that most of you REALLY care about making sure players who don't finish last get something for the effort. So much so that you don't seem to care all that much about the winner getting ELO commensurate with the difficulty OF ACTUALLY WINNING THE GAME. And, in all but scenarios so random that you have to basically make them up, I believe it is harder to win a four player game than it is a two player game (provided there's at least one highly ranked player in both games).

And, if that's the case, that's the case. I'll let it go.

The end.

[euklid314]

The 3rd player in your game only gained one single (!) ELO point. You are argueing that this player got exactly the expected position (3rd) and should not have gained ELO points. But you are wrong - he did better than expected (but only a little bit, thus he won only a single point).

I want to explain this in more detail: Even if this player had the 3rd best ELO on the table that does not mean that he is expected to get 3rd place. Elo is much more refined! By ones (relative) Elo points everybody is awarded a win percentage against each of the other players. Thus one can compute the expected probabilities for any player to get 1st, 2nd, 3rd, 4th position in a 4-player game.

If this player had, say 3% chance of winning the game, 15% chance of getting 2nd, 60% chance of getting 3rd and 22% chance of getting 4th, then his expected position is 0.03*1 + 0.12*2 + 0.60*3 + 0.25*4 = 3,07. In the long run, this player would place a little worse than 3rd place (i.e., in 100 similar 4-player-games he is expected to beat only 93 of his 300 opponents). If this player got 3rd position in a single game he was doing better than his expected value (according to ELO calculation), thus he will gain points.

If the player had had more ELO points (but still was 3rd on the table) his expected placement on the table could have been 2.6 and he had lost ELO points for getting 3rd in the game.

That being said, the ELO system is not perfect (but in my opinion superior to all your suggestions) - especially for high-luck games. In multi-player games the final result is interpreted as many 2vs2 results. The pros and cons of this approach have been discussed here extensively, it gives an incentive to fight for each position till the very end. This might be better suited to some games than others, but BGA wants a unified scoring system.
In order to avoid big swings the K-factor of multiplayer games is lowered. The amount of this reduction can be discussed but some reduction is absolutely necessary.

[Romain672]

There is another point I don't believe someone talk about.
In a specific game where there is some interactions between players:
- 2p is 'very easy' to win. Highest elo of those playing that setting are around 800.
- 3p is 'easy' to win. Highest elo of those playing that setting are around 600.

This is because at 3players, you got less control on what others players do. And it's possible that other players target the highest elo which make it harder to win.
If you just change the k-factor or something similar, that will just increase the swings. So instead of having an elo between 550 and 650, that player would get an elo between 500 and 700, and that doesn't work.

It look like, like detlefchef11 said, that the difference of elo between players should depend of the number of players.
So if that 800 elo is against that 600 in a 2p game, everything work normally.
Now if that 800 elo is against that 600 and another 400 in a 3p game, the game instead of doing two match-up 800v600 & 800v400, should reduce that difference. So you could have something (per example if you take 20%, everyone lose 20% of elo for the calculation): 640v480 & 640v320. And so as a result the highest elo player would lose less elo and win more elo when you increase the number of players.


Now, if you take the game Yahzee, that would be pretty bad since that effect doesn't happen on this game (or at least look minor). But note that adapting to another player can be done in Yahtzee, so it should have a small bonus (like maybe 5% at 3 players?).

But:
- that increase complexity of explaining elo and so increase potential bugs, especially for coop games.
- if you are doing that, you should start by doing it with the luck factor, to make high luck game have a lower k-factor and low luck game have an higher one
- the ratings of luck & interaction maybe need some changes in some games
- it can be hard to get the exact amount for that change to be benificial, but that should make the 2p vs 3p players be closer

But pretty interresting subject. Never though to that before reading that post.

[ChiefPointThief]

@ detlefchef. Yes bigger swings is a problem. This is one of the issues I have w/ arena especially as it pertains to games w/ a luck factor. It becomes more of a representation of what your success is over the last few games than the entire season.
Ex. This is a rough estimate in the change of elo of two players head to head in arena (40pts per game) compared to the 20pts in regular games.

Player A - left side is arena - right side non arena
1650
1620-1635
1598-1621
1577-1609
1558-1600
1542-1590

Player B
1530
1560-1545
1584-1559
1605-1571
1625-1582
1641-1592

In arena there is about a 230 change in elo over 5 games and player b even surpasses player a just after 3 games. Even though player a may have played tens or hundreds of games w/ an extremely higher win rate than player b.

Yes finishes outside of 1st place should be reflected in your score. If 2 Catan players play four games and player a gets 10,9,9,7 and player b scores 10,4,4,3 clearly one player is better than the other.

I invited a friend to a game of trio. Trio uses the winner takes all. They completely outplayed me that game but neither of us won. I didn’t make any matches. My friend was very confused as to how we both tied in the rankings. I told them I just live in this world I didn’t make it take it up with someone else.