Top all time ELO as a measure of a game's randomness

[smurph50]

Break the code and Can't Stop have the ELO of the best players in the 400's while other games are 700's. I hypothesized that this is related to the randomness of a game. Anyone agree or disagree?

[Poison Pink]

I sadly don't play or know these two games to be specific but my general take on it is - the all time elo is irrelevant and noone should pay attention to it or use it for any purpose because

1. It's not compiled of people playing the same game on even playing ground. The Gamemodes, Number of Players, Settings and Difficulty varies for players within the ranking. For games with multiple options you will almost always find the players on top that cut down the difficulty and remove as much variability as possible. Most commonly playing games in 2p instead of more People.

For example: its a huge difference if you play Azul in 3-4 Player versus 2 Player. 2P Azul Topdog has like 1000+ Elo which no 3-4P player will ever top because there is more variability while 2P Azul is almost a solved game.

In King of Tokyo 3p its really hard to die while in 5p arena you will regularly get eliminated and have way higher stake turns. Almost all "top all time players" in King of Tokyo play it on easy diff either 2p or 3p.

2. There is no competitive matchmakeing. You will play against anyone and everyone of any elo. So the best players in "solved games or high leverage games" will get really high elo scores but they don't mean much because the most significant amount of games they play they play against players way below their station that don't really have a shot at winning because really good players are rare and if you are top 1% player you will eat up most average to good players without breaking a sweat. Which means the bigger the playerpool the less likely to face equally good players. If there is only a small playerpool you end up being the big fish in a small pond always playing against the same players over and over.

3. There is no minimum required number of games to top the all-time-ranking -you can be listed in the top 20 with as little as 10 games.

I think the most you can say about games where best players have lower Elo numbers is that there is no safe guaranteed way to win the game over and over. If it was actually fairly measurable I would find consistantly winning in high variability games against all odds just as impressive as solved game although you will only win like 60-75% or something and not 94%

[ufm]

I think the most you can say about games where best players have lower Elo numbers is that there is no safe guaranteed way to win the game over and over.

Ironically, this means top ELO is relevant.
In games of pure randomness, a player can't win the game over and over, obviously.
In games of pure skill, skilled players can reliably beat less skilled players (unless they somehow blunder, or both players' skill levels are not so different).

Break the code and Can't Stop have the ELO of the best players in the 400's while other games are 700's. I hypothesized that this is related to the randomness of a game. Anyone agree or disagree?

You should consider the player pool size as well: less popular games tend to have lower ELO values.

[Ceaseless]

Break the code and Can't Stop have the ELO of the best players in the 400's while other games are 700's. I hypothesized that this is related to the randomness of a game. Anyone agree or disagree?

This is accurate. The highest elo of the games correlates to their skill, assuming we're comparing two BGA games to each other at least. It's one of the reasons I use the top ratings as one of my indicators on whether or not to try the game. It's not the only factor, game frequency can impact the situation, so keep an eye on if a game has been played enough to have properly reached its elos.