Page **39** of **39**

### Re: British Chess Championships 2011

Posted: **Tue Sep 13, 2011 2:29 pm**

by **David Shepherd**

The problem as I see it is that there are 11 rounds so there will be a difference in number of each colour - would it not be more sensible to have 10 or 12 round tournaments.

Anyway at the moment it seems from the above there is a bias towards the higher players having more whites. However there is an argument that you do not want the situation where the best player fails to win by half a point because they were slightly disadvantaged by having 6 blacks. A system that has a slight bias towards the better player may be fairer in terms of the final result.

### Re: British Chess Championships 2011

Posted: **Tue Sep 13, 2011 3:02 pm**

by **Ian Kingston**

David Shepherd wrote:The problem as I see it is that there are 11 rounds so there will be a difference in number of each colour - would it not be more sensible to have 10 or 12 round tournaments.

As noted by Stewart Reuben earlier, this risks some players getting a 6-4 or 7-5 split. Anyone who's ever played in a few six-round weekenders will have had a 4-2 split at some point in their career - it's quite common.

### Re: British Chess Championships 2011

Posted: **Tue Sep 13, 2011 3:32 pm**

by **David Shepherd**

Ian Kingston wrote:As noted by Stewart Reuben earlier, this risks some players getting a 6-4 or 7-5 split. Anyone who's ever played in a few six-round weekenders will have had a 4-2 split at some point in their career - it's quite common.

True, but maybe in a large competitive field combined with the fact that there are 10 or 12 rounds (compared to 6 may alter the likelyhood of this). Also the pairing system could be changed to allow for pairing of players with half a point difference in score to equalise the number of blacks and whites.

The point is however right and and maybe on reflection 11 or 9 rounds are better.

### Re: British Chess Championships 2011

Posted: **Wed Sep 14, 2011 1:10 am**

by **Stewart Reuben**

The preponderance of whites on the higher boards when the number one seed has white in round 1, should only happen if the number of players is not divisible by 4, as this year. The Acceleration was divided somewhat oddly, I don't know why. There were 23 top fraction pairings and 20 bottom fraction. One would have expected 22 and 21. Thus 12 top players got white in round 1 and only 11 got black. That set up a very small, but undesirable, bias. But much more important is that white tends to score about 53%. Thus players in the top half tend to have had more whites than blacks.

I have previously tried to overcome this bias, by giving black to every tenth higher rated player rather than white. Nobody seemed to notice that I had done this, but I didn't see any difference. Of course I didn't try that for the British Championship.

As far as I know the idea of colour equalisation taking some precedent over score in order to balance the colours in 10 or 12 rounds has never been tried. Nowadays it would be relatively easy to do a computer simulation. Much earlier in this thread, or another one, this has been discussed.

Simon Bibby's proposal that the top 10 players all get black in round 1 would not meet with approval. Nor would the higher rated gets black where two players have the same colour history (usually when pairing an odd-numbered round). Many years ago I used to alternate the alternation. Thus player 1 got white the first occasion. Next time he would get black, probably two rounds later. Another variation is that the higher rated gets his colour if above or equal to 50% and the lower rated gets his colour if below 50%.

Basically these typ of complicated rules did not meet with approval pre-computerisation because of the danger of forgetting or mixing up the rule.

The only correct way to run a chess event to eliminate colour bias is, of course, for each player to meet his opponent twice, one with each colour. Probably two games a day would then be more suitable. I can't see this becoming popular. When Hastings had an 8 player double round Premier (one round a day of course) that was ideal. As Korchnoi said, 'You have organised every player's dream'.

Stewart Reuben

### Re: British Chess Championships 2011

Posted: **Wed Sep 14, 2011 1:31 am**

by **Roger de Coverly**

Stewart Reuben wrote: There were 23 top fraction pairings and 20 bottom fraction. One would have expected 22 and 21.

The logic for that one certainly escapes me. If you accelerate, there's a reasonable argument for saying that the top half should be the even number. Simply you hope to avoid bias by giving the top half an equal spread between Black and White.

### Re: British Chess Championships 2011

Posted: **Wed Sep 14, 2011 6:28 am**

by **Alex McFarlane**

If you do a 50-50 split acceleration works when you are 'eliminating' lower half players on 100%.

When you are also eliminating bottom half players within a half point of the lead then you run the real risk of not having enough top half players to meet the bottom half. For this reason the top half should be slightly larger circa 56%.

### Re: British Chess Championships 2011

Posted: **Wed Sep 14, 2011 8:19 am**

by **Roger de Coverly**

Alex McFarlane wrote:When you are also eliminating bottom half players within a half point of the lead then you run the real risk of not having enough top half players to meet the bottom half. For this reason the top half should be slightly larger circa 56%.

Torquay 2009 had 76 players and a 19/19 split. Canterbury 2010 had 78 players and a 21/18 split. No one commented about that at the time.

### Re: British Chess Championships 2011

Posted: **Wed Sep 14, 2011 8:33 am**

by **E Michael White**

There are only two good systems for accelerated pairings:-

- Dont use one
- Pair players in the early rounds against players 100-200 points higher or lower

The irrationales behind currently used accelerated pairing systems make invalid use of probability theory and the pseudo probabilities are derived from grading systems which are not designed to produce accurate probabilities.