Author: blass uri
Date: 21:03:51 10/22/98
Go up one level in this thread
On October 22, 1998 at 22:43:35, Komputer Korner wrote: >On October 22, 1998 at 03:49:12, blass uri wrote: > >> >>On October 21, 1998 at 23:57:40, Komputer Korner wrote: >> >>>Presently there is a huge argument about alternative point scoring systems on >>>the RGCC. The one that generated the hotly debated proposal was 3 points for a >>>win and 1 point for a draw. I mistakenly assumed that the proposal was for 1/2 >>>point for a draw. My original post on RGCC is copied below. However even if you >>>substitute 1 point for the 1/2 point in the equations below, the advantage ratio >>>of white will still be larger than under the present scoring system. For proff >>>of that I repeat the equations below with the 1 point for a draw. >>>White expects (3*0.31) + 1*0.50)= 1.43 points. Black expects (3*0.19) + >>>(1*0.50)= 1.07 points. The white to black ratio becomes 1.336448 which is >>>between the new ratio below (which mistakenly assumed 1/2 point for a draw) and >>>the old present rules ratio. >> >>The probabilties are not right >>I checked the ssdf games and I found something near 30% draws. >>I do not know about 50% draws in low levels and you said that the proposal was >>not about GM games >> >> >>> >>> >>>"It will only increase the advantage of playing white which is already >>>large enough. White players enjoy a 56-44% advantage. Assuming 50% of >>>games are draws (the advantage of playing white will be even more >>>under your new system if you assume less draws), this means that the >>>spread is 50% draws, 31% white wins and 19% black wins. If you give 3 >>>points for a win and 0.5 points for a draw , >> >> >>> white's expected score >>>will be now (3*0.31) + (0.5*0.50) = 1.18 points every game. >>>Black players will now expect (3*0.19) + (0.5*.50) = 0.82 points every game. >> >>I do not agree because the players will play differently and the probabilities >>can change to 40% for white and 27% for black >> >>In this case white expected score: >>3*0.40+1*0.23=1.43 >>black expected score >>3*0.27+1*0.23=1.04 >> >>Uri > > >It doesn't matter what % draws there are as long as white's overall score is >greater than 50% and this is true for any ELOs above 1800. As long as white's >overall score is greater than 50% the equations hold for any draw % you want to >plug in. Your last comment still backs up the conclusion that white's expected >score is higher than black's. Therefore if one player gets a greater number of >whites in a swiss which is already the case, then any proposal to award wins by >a bonus over the present system of awarding 1 point for a win will skew the >white winning ratio higher. Indeed in your last calculations the figures are >wrong. They don't add up to 100%. The last term of each equation has to be 0.50 >to represent the draw. Even so you still show 1.43/1.04 which is a higher ratio >than the present system. I agree that my calculation was wrong The probability for a draw should be 33% and not 23% if the probability for white is 40% and for black is 27% so the ratio should be 1.53/1.14 It is higher than the present system so if we want to change the system it is better to give black more for a draw then white. Uri > >-- >Komputer Korner
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.