The ELO (rating) system is far, far too fluid.

[KDLadage]

I have been playing chess for years. I play a lot of rated games and my rating, although it fluctuates a few points up and down, does not make wild swings of 20+ points in a single game.

The ELO system on BGA is so fluid as to be meaningless.

Let's take a look at my rating for Azul. Not long ago, I broke 200. Then 300. Then came very close to 400, dropped down to below 300. Then back up. then down. And so on.

The amount by which the ratings in the games is adjusted by a single game is far too high! Am I a low-to-mid 200 player? Or am I a player at nearly 400? That all depends upon which day of the week (and which hour in that day) we are discussing.

That system needs an overhaul if it is to have any meaning.

[ufm]

https://boardgamearena.com/faq?anchor=faq_account_elo

In short, the variance is larger because you play multiplayer games.

[KDLadage]

I have read that -- and I understand what they are saying.

I am saying that I disagree with the fluidity. I sincerely believe it needs to be reigned in. A lot.

[nik592]

You're also discounting the luck factor in games other than chess - not that there aren't other games that are pure strategy, but chess certainly is, and Azul certainly is not. The more luck is in a game (and Azul isn't a huge amount, but still), the bigger swings that will occur. As I understand it, ELO wasn't designed to take into account luck or multi-player scenarios, so it doesn't fit well here. I believe there are other rating systems and I don't know if they handle those factors better? But as we've recently seen from BGA's tinkering with ELO, that can lead to even worse situations, where ELO is even more meaningless and frustrating. So I'd wonder about the advisability of BGA trying to tinker with it to make multi-player games better on ELO (while still also maintaining the system for 2p), let alone trying to adjust for luck (which would be even more difficult).

[KDLadage]

"You're also discounting the luck factor in games other than chess" -- I am not discounting anything (at least I do not believe I am). I just think that if my rating, on a scale of 0 to less than 1000, can swing by ~200 points in a few days, then it is a meaningless measure.

===

Imagine a system that worked something like this: At the end of the game, start with an adjustment of +0.

If you have a score in the game that is better than at least one of your opponents, add +2 points of adjustment. Check the highest ELO of those that you have beaten in the game. Add +1 point to the adjustment for every full 50 points of their ELO above yours.

Next, if you have a score in the game that is worse than at least one of your opponents, add -2 points to the adjustment. Check the lowest ELO of those that you have lost to in the game. Add -1 point to the adjustment for every full 50 points of their ELO below yours.

At least that would be far less swingy.

Just a thought.

[Romain672]

I suggest you that post I created: https://boardgamearena.com/forum/viewtopic.php?t=38068 (skip my first post, and directly read the answers).

[RobertBr]

Its a legitimate point.

The problem is how you get players to something nearly their true skill level and how you then adjust when their skill level changes.

Bear in mind before I played my first rated game of Chess I had read several books and played hundreds of games. That is more than for almost any of the types of games on BGA. So my Chess level was going to be much more stable once I started playing.

[FSKFSK]

Elo does not work for games with a luck component. For some games on this site, someone who knows the basic strategy will win at least 30% of the time (Jump Drive, Abandon All Artichokes, many others). For a game like that, the Elo system will lead to an unfair result when one player has a much higher rating. You will lose Elo on average playing someone with an elo 100+ points lower than yours, because the chance Elo gives you of winning is higher than your actual chance.

The correct solution is to use a different probability distribution than Elo for games with a significant luck factor. For example, Elo as implemented here caps out where the better player has a 90% chance of winning. (You always get +2 for a win, no matter how much lower rated the loser is.) For some games, the cap should be 70% or even 55%.

[KDLadage]

I believe (and I could be wrong) that those who are focusing on the fact that some (most?) games have elements of luck, kingmaking, and the like.

That is not the issue (best I can tell).

The issue is that the calculations for ELO adjustment are so fluid as to exacerbate the issue of luck, kingmaking, and so on.

If the ELO adjustment calculations were not as volatile as they are, this would not be such an issue. So reign them in.

My suggestion above might be too extreme for some. So what about something like this: if your ELO score at the start of a game is less than 200, then do not change anything. If it is 200-299, then divide all adjustments by 2 and round toward 0. If it is 300-399, then divide all adjustments by 3 and round toward 0. If it is 400-499, then divide all adjustments by 4 and round toward 0. And so on...

Just something -- anything -- to stabilize it a bit. Make it harder to reach each +100 points, and stabilize things when you get there.

Anyway. This is just my thoughts. I can see that there are a lot of thoughts when it comes to how the ELO is calculated, and they are far from a consensus. That, and I really feel that the discussion of the luck factor for this or that game completely misses the point -- light years away from the issue I am trying to discuss (if anything, high luck games should reduce the ELO shifts, not exacerbate them).

But nobody seems to want to discuss what I am trying to discuss. So I will relax and say no more on this topic.

[Romain672]

Azul have 3/5 luck, one example I know very well is Legendary inventors which got 1/5 luck, chess that you mentionned got 0/5.

Thing is, I was happy with the swing of elo of legendary inventors. One of the goal of elo is to have fun too, seeing a number going slowly up because the k-factor is really low doesn't look fun.
So yeah I was going from 750 to 850 elo depending of my victories and defeats, but that was good to me.



(edit: it could be because I was 1st~3rd top in elo. Maybe if I was like 10th or 20th that would be completely different)