Upgrade to Chess.com Premium!

Chess rating system


  • 3 years ago · Quote · #1081

    andreic

    I have an idea about how to rate players during a RR tournament., and a general extension in mind.

    I believe a measure of the player strength relative to a certain game is given by the speed of the material gain. At the begining the material is even, and at the end  there are two posibilities:

    1. The game is a draw(any chess rule leading to a draw) .  

    No one loses both sides keeps its armies for a feature battle. In this case the game score is given by

    (material1 - material2)/(number_of_game_moves). One of the sides will have a positive score , while the other will have the same absolute score but with the sign changed.

    2. The game ends in mate. The side which loses the game is considered to lose all  the material in the end, as the soldiers cannot fight without their king.

    The score will be for the winner: 

    1.material_of_the_winner_at_last_move/number_of_game_moves

    And for the loser will be 

     (- material_of_the_winner_at_last_move)/number_of_game_moves.

    At a RR tournament is sufficient to add the score for every game for every player just like that.

    If a player competes in different tournaments we must take into account the adversary strength, and the player ranking  is computed different.

    The player strength is the average  of the cumulated game scores.

    Suppose the players strengths are s1 and s2,number of games played so far are n1 and n2, and player 1 has a positive game result: r( a draw does not need to be a zero result).

    Then compute

                        s1 =  (s1*n1+(s2/(s1+s2))*r)/(n1+1);

                        n1=n1+1;

                        s2=(s2*n2-(s2/(s1+s2))*r)/(n2+1);

                        n2=n2+1; 

    A simple numerical example s1=0.5 s2=0.26  n1=10 ; n2=25; r=1.25

     New stregths:

      s1= (0.5*10 + (0.26/0.76)*1.25)/11 = 0.49

      s2= (0.26*25 - (0.26/0.76)*1.25)/26 =0.23

    Also the probability the player 1 wins the player 2 is computed like (s1/s1+s2).

    Before the game p=65%; after the game p=68%.

    The interpretation of score 1.25:

    If the number of moves in current game was 30 , it means the winner had 30*1.25 in material at the last move.

    I will apreciate some feedback and how can we test this. 

     

    Andrei from Romania

  • 3 years ago · Quote · #1082

    browni3141

    Andrei, what if you sacrificed material before mate? Your rating gain would not be as great yet a win through material sacrifice could be considered more skillful than a win retaining material.

  • 3 years ago · Quote · #1083

    andreic

    browni3141 wrote:

    Andrei, what if you sacrificed material before mate? Your rating gain would not be as great yet a win through material sacrifice could be considered more skillful than a win retaining material.


    There is a compensation for that. If the sacrifice is not seen by the opponent the game ends quickly and the ratio material/number_of_moves is still high.

    Also keep in mind that in this situation the material of the winner will be a big number because the opponent material does not count after mate(it is considered captured by the mate move):

    Suppose after a few moves one sacrifices it's queen and then mate. His material

    is a bigger number compared to the situation when no sacrifice is done and the game lasts for 40 moves.

  • 3 years ago · Quote · #1084

    Samsch

    Hi.

  • 3 years ago · Quote · #1085

    Rafchess

    InnocentIts somr thing !! let me take little respite. then resume.

  • 3 years ago · Quote · #1086

    andreic

    To follow the basic idea of my system a positive score of 1.25 means the winner, takes 1.25 points per move on average from opponent.Remember that the mate 

    move "captures" all opponent pieces.

  • 3 years ago · Quote · #1087

    tubebender

    Don`t try to" reinvent the Wheel"; stay with either the Elo (for simplicity) or the Glickman modified Elo system used by the USCF.

  • 3 years ago · Quote · #1088

    Rafchess

    Innocenttubebender wrote:

    Don`t try to" reinvent the Wheel"; stay with either the Elo (for simplicity) or the Glickman modified Elo system used by the USCF.


     Innocent more bending will spoil the tube.so appriciate ur understanding!

  • 3 years ago · Quote · #1089

    cathen

    Merry Christmas everyone!

  • 3 years ago · Quote · #1090

    GPGP

    thanks, same to you and everyone

  • 3 years ago · Quote · #1091

    Rafchess

    Thnx ! same  to  everybody.Laughing

  • 3 years ago · Quote · #1092

    chestonblumentahl

    anyone else get a little turned on by the bishop?

  • 3 years ago · Quote · #1093

    Rafchess

    chestonblumentahl wrote:

    anyone else get a little turned on by the bishop?


     Frown...unpredictable !!

  • 3 years ago · Quote · #1094

    Rafchess

    andreic wrote:

    I have an idea about how to rate players during a RR tournament., and a general extension in mind.

    I believe a measure of the player strength relative to a certain game is given by the speed of the material gain. At the begining the material is even, and at the end  there are two posibilities:

    1. The game is a draw(any chess rule leading to a draw) .  

    No one loses both sides keeps its armies for a feature battle. In this case the game score is given by

    (material1 - material2)/(number_of_game_moves). One of the sides will have a positive score , while the other will have the same absolute score but with the sign changed.

    2. The game ends in mate. The side which loses the game is considered to lose all  the material in the end, as the soldiers cannot fight without their king.

    The score will be for the winner: 

    1.material_of_the_winner_at_last_move/number_of_game_moves

    And for the loser will be 

     (- material_of_the_winner_at_last_move)/number_of_game_moves.

    At a RR tournament is sufficient to add the score for every game for every player just like that.

    If a player competes in different tournaments we must take into account the adversary strength, and the player ranking  is computed different.

    The player strength is the average  of the cumulated game scores.

    Suppose the players strengths are s1 and s2,number of games played so far are n1 and n2, and player 1 has a positive game result: r( a draw does not need to be a zero result).

    Then compute

                        s1 =  (s1*n1+(s2/(s1+s2))*r)/(n1+1);

                        n1=n1+1;

                        s2=(s2*n2-(s2/(s1+s2))*r)/(n2+1);

                        n2=n2+1; 

    A simple numerical example s1=0.5 s2=0.26  n1=10 ; n2=25; r=1.25

     New stregths:

      s1= (0.5*10 + (0.26/0.76)*1.25)/11 = 0.49

      s2= (0.26*25 - (0.26/0.76)*1.25)/26 =0.23

    Also the probability the player 1 wins the player 2 is computed like (s1/s1+s2).

    Before the game p=65%; after the game p=68%.

    The interpretation of score 1.25:

    If the number of moves in current game was 30 , it means the winner had 30*1.25 in material at the last move.

    I will apreciate some feedback and how can we test this. 

     

    Andrei from Romania


     Laughing appriciate ur Romanian formulae!! Wish u good luck % Happy New Year!

  • 3 years ago · Quote · #1095

    andreic

    Thanks.  A good year for every one.

  • 3 years ago · Quote · #1096

    Rafchess

    andreic wrote:

    Thanks.  A good year for every one.


     Smilewish the same to u & all in Chess.com !!

  • 3 years ago · Quote · #1097

    icebergslimshadow

    [COMMENT DELETED]
  • 3 years ago · Quote · #1098

    beardogjones

    "points" is an artificial concept - why not just throw in the towel and

    have Houdini assign the ratings or play both sides for that matter?

  • 3 years ago · Quote · #1099

    Rafchess

    beardogjones wrote:

    "points" is an artificial concept - why not just throw in the towel and

    have Houdini assign the ratings or play both sides for that matter?


     Innocent that could be an wonderful innovation to ur artificial concept!! of points.happy new year beardogjones.

  • 3 years ago · Quote · #1100

    Burginflickle

    [COMMENT DELETED]

Back to Top

Post your reply: