Board Game Arena

• There,
• there
• and there

• Players having high ELO with previous formula will keep having high ELO because it's way more difficult to gain&loose points with new formula.
• Players who loose an important number of points during the last period of time will have difficulties to restore their rankings.
• The new formula does not give enough points - which is not really encouraging.

• With the old formula, when 2 players with the same ELO plays together, they gain/lost +30/-30 points.
• With the new formula, when 2 players with the same ELO plays together, they gain/lost +15/-15 points if they played less than 30 games, and +7.5/-7.5 afterwards.

• More accurate rating system. By using the ELO formula, the rating difference between 2 players determine their probability to win.
• The new players rating converge to their "real" value more faster, and are more "stable" afterwards.

• We consider that each player win a 2-players game against each opponent after him in the game result, and loose a 2-players game against each opponent before him in the game result.
• This way, a N-players game is equivalent to (N.(N-1)/2) 2-players games.
• Of course, the ELO difference is divided by (N-1) because there is no reason than a game with a lot of players makes the ELO ranking progressing faster than a 2-players game.

• Find with you what is the best "elasticity" for ELO (like before ? less than before ?).
• Recalculate ELO rankings, taking into account all games from the beginning of BGA (yes, this is feasible).
• This way, everyone will get an ELO ranking as if we always used this new formula in the past.

More accurate rating system. By using the ELO formula, the rating difference between 2 players determine their probability to win.

Win %       New system       Old system 
90%            366             752
80%            240             564
70%            149             376
60%            72              188
50%            0               0

Players having high ELO with previous formula will keep having high ELO because it's way more difficult to gain&loose points with new formula

How do we address the "N-players" games with N>2:
We consider that each player win a 2-players game against each opponent after him in the game result, and loose a 2-players game against each opponent before him in the game result.
This way, a N-players game is equivalent to (N.(N-1)/2) 2-players games.
Of course, the ELO difference is divided by (N-1) because there is no reason than a game with a lot of players makes the ELO ranking progressing faster than a 2-players game.

Find with you what is the best "elasticity" for ELO (like before ? less than before ?).

• K=60 for the first 30 games (= same elasticity as before).
• K=30 afterwards (= +15/-15 points per match for a 2p-game)

• For 2p-games: K=60 for the first 30 games, K=30 afterwards.
• For 3p-games and more: K=75 for the first 30 games, K=45 afterwards

sourisdudesert wrote:

• K=60 for the first 30 games (= same elasticity as before).
• K=30 afterwards (= +15/-15 points per match for a 2p-game)

sourisdudesert wrote:Another additional proposal: we can introduce an elasticity bonus for game with more than 2 players, in order to take into account the fact that these games are more time-consuming and to encourage this type of game. Full example:

• For 2p-games: K=60 for the first 30 games, K=30 afterwards.
• For 3p-games and more: K=75 for the first 30 games, K=45 afterwards