Board Game Arena

Pistolero wrote: ↑03 May 2019, 12:24 Each game should have a default discrepancy

So, just to verify: in order to implement the suggestion, statistical analysis needs to be done on all 100+ games already on BGA, and each new game (before it is released) needs to have statistical analysis of what the normal discrepancy is?

Pistolero wrote: ↑03 May 2019, 12:24 For chess, ELO works fine, so you have your answer

So there's no such thing as a close chess game, or a blowout win? Or would ELO calculation be inherently different for different games on the site?

As I posted in another thread, I tend to play devil's advocate. The suggestion may have merit, but please look at the implications of implementation.

Jest Phulin wrote: ↑04 May 2019, 00:30 So, just to verify: in order to implement the suggestion, statistical analysis needs to be done on all 100+ games already on BGA, and each new game (before it is released) needs to have statistical analysis of what the normal discrepancy is?

And all that assuming each game has exactly one possible way of ending / scoring the match, and that the end result from these scores accurately reflect how close each player was to winning. Off the top of my head, I can name Innovation and Lords of Xidit as counterexamples for these respective assumptions, and I'm fairly certain there are many more.

So yeah, while I appreciate Pistolero's enthusiasm and optimism w/r/t "easily" creating a superior formula, there may be a few unexpected hiccups along the way – thanks for (usually) stepping in with on-point questions when The People demand something akin to these bold claims be implemented without exercising their due diligence

I agree with the idea from pistolero :

Once you reach the top, you disable ELO rank based forever to stay at the top level. I think all players should lose a very small amount of ELO points every day (like 1 per day, and only if you don't play ELO enabled games)

Look at Hex : I have never played against players > 300 for more than 2 years, and I am still ranked 568 !!

A the opposite, at Blooms, I have won 28 of my 30 latest games, I have defeated the player ranked #1 (1v / 1def), and the player ranked "4 (3 victory / 1 def) and I am only ranked #8 at thad day.

In summary, I think the Elo system is too conservative.

Cedroc92 wrote: ↑18 May 2019, 21:55 Look at Hex : I have never played against players > 300 for more than 2 years, and I am still ranked 568 !!

A the opposite, at Blooms, I have won 28 of my 30 latest games, I have defeated the player ranked #1 (1v / 1def), and the player ranked "4 (3 victory / 1 def) and I am only ranked #8 at thad day.

I looked at the Blooms-ranking and had a look at the player stats of the players who are before you in the list. I didn't analyze it too deeply but to me it looked like the players who are before you in ranking had even more impressive stats than you do (more games, higher win-%, also playing very strong other players) - so I don't see what's wrong with them being placed on the top. Just keep playing... as you're still on your way up (probably haven't reached your top yet) I think you can still gain more ELO-points and eventually climb the top.

As for Hex: I think if you played so well that you reached 568 ELO I think your skill is very high. Do you think you lost a lot of your skill within these past 2 years? Or why are you surprised you still have those ELO points?

Remember: ELO is not supposed to be a ranking of aktive players but it's supposed to be a measure of skill in the game. In contrast to physic sports like Tennis etc. I don't believe the loss of skill due to lack of practice or due to aging is very relevant in boardgaming so I don't see why ELO should be lowered because of inactivity.

I understand people want a ranking-system of active players and right now that exists and is called "leaderboard" and is derived from ELO with just a little change (people without game within 2 month not shown). Yeah, I also think there should be a change on leaderboard (maybe still based on ELO but with a lot more modifications considering activity and maybe chosen opponents level) - but I think the admins already work on it as they stated in another thread.
But that has nothing to do with ELO which IMO should stay a measure of skill independent from activity.