Elo count

[Sombracier]

We don't "care" if the elo are limited to 5000 500 or 50 points, the concern is how it behaves, related to how it used to behave.

Well some of you might think nothing has changed recently i will keep an open eye on this

[Stroom]

You are just salty that your elo is not increasing any more. This is your current skill level and if you want it to raise, you have to learn more and start playing better with more complex settings. That's all there is. By having a loss deduct more points than winning adds is the only thing that makes sure that you will not indefinitely increase the elo score. You have to play better than that.

[Sombracier]

For the last time, I don't answer for people but i'm not "salty" or whatever. If my games leads to a lower score, so be it. This wasn't the point so please stop posting nonsense that lead the reader away from the core topic which was "the point algorithm seems/seemed to behave differently recently, is it real or just an illusion ?"

[Stroom]

I do not see it. This is probably due to the complainers playing with higher average elo games so that's why they think this way.

[Jellby]

for example:
I play a 3-way game (me 560 elo, my friends 260 and 960 elo points). We take 6 colors and black powder (thus difficult game and with an average player).
- if we finish everything, we have +4
- if we finish but with 2 or 3 cards not put, we have -7

You can find the "bot's ELO" at viewtopic.php?t=4814

With 6 colors and 3 players, 28 points is 505, 29 points is 600, 30 points is 720. If the team's average ELO is below the bot's ELO, you'll increase it, if it's above, you'll decrease it.

So in you case, you have an average of 593, so anything below 29 points will decrease your ELO, and 29 points will hardly change it.

[Jellby]

By the way, if you want to figure out what's the bot's ELO for some particular setting/score, after you finish the game do the following:

1. Take the ELO change, preferably with 2 decimal places (which you get by hovering on your ELO in the result screen). Call this x
2. Take the average ELO of the team, which is reported in the result screen too. Call this y.
3. Assume your K factor is 20, which should be true if you have played more than a handful of games.

Now compute:

y + 400 × log [ (K/2 + x) / (K/2 - x) ]

where "log" is the base-10 logarithm.

So take for example https://boardgamearena.com/table?table=289957046

x = 752.62 - 760.98 = -8.36
y = 760
K = 20

(20/2 + (-8.36)) / (20/2 - (-8.36)) = 0.089325
log(0.089325) = -1.049029
760 + 400 × (-1.049029) = 340

and sure enough, the table says that for 3 players, 6 colors and a score of 26 the bot's ELO is 340. This tells us that the flamboyant 5s option (which was on in that game) didn't affect the ELO computation.

[ChiefPointThief]

Specifically the 55 cards version is completely outside of elo-calculation. Playing that setting will never let you gain/lose any elo points. I think because it was considered too luck-dependent?
Any other color-combination (5 colors / 6 colors / multicolor / with or without black) will have elo-adjustment. Only this one doesn't.

I don't know why there's no hint about that when creating a table. Anyway - now you know

If one of the low multicolor cards are on the bottom of the deck it is impossible to get a perfect score. Couldn't this be mitigated by allowing the variant in which players are allowed to finish playing out their cards in hand once the deck is exhausted? Then 55 card games could be ranked.

[Stroom]

This happens so rarely that it does not affect your elo score very much. Just play more games and your score will adjust itself. Over hundreds of games, it is negligible. You can just consider an occasional elo loss as a part of the elo system. No big deal.

[Wreckage]

The 55 card version is indeed the most difficult to get high scores. Currently you always gain 0 ELO no matter if you get 30, or bomb out, or any score in between.

If a dev decides to adjust ELO calculations that include the 55 card version, they should keep in mind that keeping a 26 score average is about equivalent to keeping a 29 average on avalanche. Not only is there the possibility of a multi card near the bottom, making perfect scores impossible, but the deck is also shorter. You must play the same number of cards, while having 5 fewer tokens during the game, and 5 fewer safe discards. In fact, it's probably harder without black powder, if you have good conventions for BP.

I agree with ChiefPointThief that all the variant's in the rules should be allowed. Of course, if you play with the variant allowing players to continue after the deck is exhausted that makes a high score much more likely, so the ELO should be strongly penalized if that option is turned on.