Estimating White's Advantage

General discussions about grading.
E Michael White
Posts: 1301
Joined: Fri Jun 01, 2007 6:31 pm

Re: Estimating White's Advantage

Post by E Michael White » Mon Aug 17, 2015 10:00 am

IM Jack Rudd wrote:
E Michael White wrote:My line of thinking was that under CAA pairings where a tournament has an odd number of boards, say 21 then the stronger player out of the two on each board in the first round is more likely to be white. In the case of 21 boards, 11 of the players will be white and likely stronger than their opps whereas only 10 black players could be stronger.
The official rules for CAA pairings state that the top seed's round 1 colour is chosen randomly, which rather demolishes that argument.
Currently almost. Historically no.

NickFaulks
Posts: 4458
Joined: Sat Jan 02, 2010 1:28 pm

Re: Estimating White's Advantage

Post by NickFaulks » Mon Aug 17, 2015 10:53 am

I have looked at my own results from 2014/15 and found little of interest regarding colours. However, I also reviewed them as regards results against players of various ratings, which is another of Ian's interests.

Against players rated 160 plus, ranging up to 202, I scored +12 =8 -12. Against those in the range 120-159 I scored +7 =5 -9. This suggests that in my case at least, and 53 games is not a tiny sample, the whole exercise is pointless since the grading system has no predictive power whatsoever.

MartinCarpenter
Posts: 2401
Joined: Tue May 24, 2011 10:58 am

Re: Estimating White's Advantage

Post by MartinCarpenter » Mon Aug 17, 2015 2:19 pm

Worth doing if you do it for a whole team though :) Those sorts of states would get you stuck on as a high a board as sanely legal.....

Howard Grist
Posts: 82
Joined: Tue Dec 12, 2006 1:14 pm
Location: Southend-on-Sea
Contact:

Re: Estimating White's Advantage

Post by Howard Grist » Tue Aug 18, 2015 12:54 pm

Analysis of the ECF grading database reveals, for games in which colours were given:

White wins - 384269 (40.83%)
Draws - 217610 (23.12%)
Black wins - 339235 (36.05%)
Total - 941114

So White score 52.39% in total.
ECF Grading System Programmer

E Michael White
Posts: 1301
Joined: Fri Jun 01, 2007 6:31 pm

Re: Estimating White's Advantage

Post by E Michael White » Wed Aug 19, 2015 10:44 am

Howard Grist wrote:Analysis of the ECF grading database reveals, for games in which colours were given:

White wins - 384269 (40.83%)
Draws - 217610 (23.12%)
Black wins - 339235 (36.05%)
Total - 941114

So White score 52.39% in total.

In the British Championships, championship section the FIDE rating difference figs for 2015, 2014, 2011 and 2010 (2013 and 2012 unavailable) are:-

Code: Select all

Year      Average	Average	Average	
		    White     Black  	Rating 		
		    Rating	 Rating 	Difference
		    
2015		2152		2148		4		
2014		2183		2179		4		
2011		2240		2232		8		
2010		2226		2224		2
Arbiter conduct in tournaments can contribute to the better performance of white causing bias of up to about 10% of the increased performance. Just as most of us are right handed it may be that arbiters concentrate on white seekers more than black seekers when looking at stronger players.

David Sedgwick
Posts: 3113
Joined: Mon Apr 09, 2007 5:56 pm
Location: Croydon

Re: Estimating White's Advantage

Post by David Sedgwick » Wed Aug 19, 2015 11:06 am

E Michael White wrote:In the British Championships, championship section the FIDE rating difference figs for 2015, 2014, 2011 and 2010 (2013 and 2012 unavailable) are:-

Code: Select all

Year      Average	Average	Average	
		    White     Black  	Rating 		
		    Rating	 Rating 	Difference
		    
2015		2152		2148		4		
2014		2183		2179		4		
2011		2240		2232		8		
2010		2226		2224		2
Arbiter conduct in tournaments can contribute to the better performance of white causing bias of up to about 10% of the increased performance. Just as most of us are right handed it may be that arbiters concentrate on white seekers more than black seekers when looking at stronger players.
I have to accept that that is an interesting finding.

Would you be able to do the same exercise for a tournament which uses computer rather than manual pairings?

If there is a pro-White bias, that should help to establish whether it's caused by arbiter performance, as you suggest, or by some far from obvious defect in the Swiss Pairing Regulations.

NickFaulks
Posts: 4458
Joined: Sat Jan 02, 2010 1:28 pm

Re: Estimating White's Advantage

Post by NickFaulks » Wed Aug 19, 2015 11:27 am

David Sedgwick wrote: I have to accept that that is an interesting finding.
Would you be able to do the same exercise for a tournament which uses computer rather than manual pairings?
I don't think that any of these figures has strong statistical significance in itself - the 8 may be borderline. However, it is difficult to ignore the fact that they are all positive. Analysis of other events is clearly needed.

E Michael White
Posts: 1301
Joined: Fri Jun 01, 2007 6:31 pm

Re: Estimating White's Advantage

Post by E Michael White » Thu Aug 27, 2015 10:21 am

David Sedgwick wrote:Would you be able to do the same exercise for a tournament which uses computer rather than manual pairings? .
I’m afraid I can’t produce that information easily, as I do not know exactly which events use computer or manual pairings.
David Sedgwick wrote:If there is a pro-White bias, that should help to establish whether it's caused by arbiter performance, as you suggest, or by some far from obvious defect in the Swiss Pairing Regulations.
This pro white stronger player bias has existed for 30-40 years. I wouldn’t be surprised if the reasons varied over time.

My initial post was to show that a straight assessment of White’s advantage by looking at the overall result percentage is dubious. I did not want to become involved in an assessment of arbiter performance as you suggest but chose the British Championship as games are readily available and all ratings shown; some may be ECF conversions. I expect arbiters conduct the pairing exercise in the British Championships by following the regulations carefully.

The following information may be of interest. I can’t completely guarantee accuracy of the Chessbase10 analysis icon or the collected data. However if anyone else wants to check individual tournaments it is relatively easy.

Recent TWICS, these will be mostly tournaments with some team matches.

Code: Select all

Year	File		game		average		average		average				White	
			count	  	White	 	 Black		difference			%
					rating		 rating					  	score

2015	TWICS1081	5557		2162		2157		5				54	
2015	TWICS1082	5606		2123		2115		8				54
2015	TWICS1083	4136		2186		2183		3				54
2015	TWICS1084	4140		2176		2172		4				54
2015	TWICS1085	4313		2169		2161		8				54

Data source:  TWICS
Data Analysis: Chessbase10 statistics icon
Last edited by E Michael White on Sun Apr 08, 2018 10:53 pm, edited 1 time in total.

NickFaulks
Posts: 4458
Joined: Sat Jan 02, 2010 1:28 pm

Re: Estimating White's Advantage

Post by NickFaulks » Thu Aug 27, 2015 10:45 am

If the numbers in the box above really are as they appear ( I don't know why they wouldn't be, but an independent check is always valuable ), they are quite startling. The FIDE Commission overseeing Swiss pairings should certainly take a look.

Post Reply