Elo rating calculation needs new statistics applied

[DancingHeart]

So, this last game I played, Takenoko, I came in last; the point totals were: 39, 34, 33, and my 32 respectivly on game table 36194798. My 32 wasn’t a huge point loss compared to the two previous players. However, the ugly thing is that my ELO on this game dropped by 75 points! This isn’t the first time a huge drop has happened on a game for me, but I have not seen such a drop with other players, either. One thing is certain, i will never donate another $20 to the BGA again, and I have selectively stopped playing games, and will continue to do so with this bogus system of scoring rankings.

[Romain672]

Hello,
Your point have no incidence. Only your ranking is important.

You loose many elo because of two factor :
- 4 players games : if you finish last, more there is player, more you will loose elo.
- Number of game played : While you have less than 20 games played on one game, you will loose and gain many elo (i think it's x3, x2 between 20 and 40games, and x1 after).

Note your last three game give you : +0 (2nd/3), +45 (1st/3), and +69 (1st/4) elo.
(and your +69 was on a game with these score : 32/32/31/23, so close too)

[jinping]

the 'best player' in Takenoko davidmieal uses the 'play-only-against beginners strategy'. you can learn from that

[Shaq Jenkins]

Many players (myself included) have complained for years about the broken ELO system, as well as the ridiculous ranking/trophy system. ELO works fine for 2 player games, which is how it was originally designed. However, ELO is TERRIBLE for 3+ player games for several reasons, and it is extremely inefficient to apply this system to games that it was never intended for.

Another problem with ELO is that it works well for some games but poorly for others. It is great for strategy games that don't have much luck, such as Puerto Rico, Tash-Kalar, and Madeira. It is not accurate for Love Letter, Can't Stop, or Secret Moon, which are games based largely on luck.

OP, I see that you lost 75 points by coming in fourth place in a Takenoko game. Honestly, 4 player Takenoko is a huge crapshoot and too hard to manage, which is why I don't play it. 2 player Takenoko is a great, complex strategical game that doesn't focus too much on luck. Your strategy changes based on how many people are playing - I don't approach 4 player Takenoko the same way as 2 player Takenoko. They're essentially different games, so why are they being scored the same way?

ELO is supposed to make it easier to find similarly skilled opponents, but how can we if it doesn't accurately measure a person's skill level? Just because someone has a high ranking when playing 2 player Takenoko against noobs doesn't mean they're good at playing 3 player Takenoko against other Experts or Masters.

This is probably too much to ask, but BGA needs to create a statistical scoring system for each individual game, while taking into account how many people are playing. The least they could do is create a system for different game genres (Saboteur and Secret Moon could be scored similarly, just like Puerto Rico and Chess could be).

[Liallan]

My first comment to this is the same one I always make: that I've never seen a skill system yet that people did not complain about. They use Microsoft's TrueSkill system at Yucata and people complain about it too, and it's even more funny when one of those complaints wants them to change it to ELO - and then over here people complain about ELO! At Yucata they even have an adjustment for categories of luck, and that doesn't stop the complaints. I don't think they could make a system that would prevent people from complaining about it. (By the way, I say the same thing over there.)

Having to adjust for all the various games, the luck levels of them, and separate player counts, and further adjust luck levels for different player counts.... sounds like a total nightmare.

But as to OP's original comment.... your score in the game has nothing to do with it. I have to agree with Romain679. You gained 69 points in a game you won by a tie breaker for Pete's sake. You gained that many for the same reason: score isn't relevant, you beat 3 other people, and you get bigger changes (either direction) when you don't have many games. I don't see you complaining about that! You gain 69 in a tie breaker and you'll pay them, but if you lose 75 for last place in a close game, you stop paying them. That's some pretty weird economics.

[sparr]

it's even more funny when one of those complaints wants them to change it to ELO - and then over here people complain about ELO!

To be clear, the system used here is not actual ELO. The people who want to switch to ELO are asking for actual ELO, not a switch to BGA's system.

[Liallan]

it's even more funny when one of those complaints wants them to change it to ELO - and then over here people complain about ELO!

To be clear, the system used here is not actual ELO. The people who want to switch to ELO are asking for actual ELO, not a switch to BGA's system.

Often the people wanting the ELO are people who play here and yes are asking for BGA's system. (Not to mention that what they used to do here more closely resembled ELO than what they do now. And the fact that ELO was meant for 2 player will always be true, regardless who is doing the asking.)

The real point of that comment was that the grass is always greener on the other side.

[DonRoger]

No system is perfect, but I would appreciate a special system for Hanabi and any other cooperative games that may arrive, one in which you get more rather than fewer points the more players you have to deal with.