Check out Janggi (Korean Chess), our featured variant for December, 2024.


[ Help | Earliest Comments | Latest Comments ]
[ List All Subjects of Discussion | Create New Subject of Discussion ]
[ List Earliest Comments Only For Pages | Games | Rated Pages | Rated Games | Subjects of Discussion ]

Single Comment

Asymmetric Chess. Chess with alternative units but classical types and mechanics. (8x8, Cells: 64) [All Comments] [Add Comment or Rating]
💡📝Dmitry Eskin wrote on Thu, Jan 7, 2021 06:51 AM UTC in reply to Dmitry Eskin from Fri Dec 16 2016 05:17 PM:

Update rules.

Remove next rule:

  1. Elvish Knights and Elvish Bishops are exchanged with starting places: Knights stay on the files "c", "f" and Bishops stay on the files "b", "g" (need to defend all Elvish Pawns)

This rule was introduced because of Elf-Elf mirror matchup, having 1. Qc3 move attacking g7 Pawn (and opponent must reply 1.... gf6 or ef6, because if Qe6 then 2. Bc4!).

But this rule has influence to Elf-Human matchup, having 1. Rh4 attacking h7 pawn (and if 1.... g6 or Nf6 then 2. Bd4).

So I think this questionable rule must be cancelled.

Now I start to get new statistics of mathups, with Fairy-Max auto battles:

  • at 1000 ELO (0.6 sec per 40 moves)
  • at 1400 ELO (6 sec per 40 moves)
  • at 1800 ELO (1 min per 40 moves)
  • at 2000 ELO (10 min per 40 moves)

How I get these ELO? I have ~95% winrate FIDE chess White vs Black when 1st White engine has 10 times more time than 2nd Black engine.

At low ELO I can get statistics very quick and modify rules to balance sides for it. But at higher ELO I can get statistics much slower, and can just check if balance is OK or not.

~10 000 games must be played for 1% accuracy (~100 games has only 10% accuracy - even equal sides may play 60-40).

I think the only things to be modified must be promotion rules. For example, introducing limitation of Orcish Pawns to promote only minor Pieces, lower Orcish winrate from 58% to 52%. Now I think that Elves and Orcs are a little stronger than orthodox Humans (~+2% winrate) and this point must be fixed.

Also, the most interesting thing is getting Pawns and Pieces values (at different ELOs). By statistics and changing start Pieces I can get only starting values. For example, starting values of orthodox Knight and Bishop are equal.

But there is also mid values which can be used at middle-game exchanges (theoretically, these mid values are best values for Fairy-Max engine, i.e. with these values Fairy-Max must have maximal winrate (vs engines with other values). For example, mid values of orthodox Bishop will be greater than orthodox Knight, because having less Pieces and Pawns at the board than at start.