Hastings Results

[IM Jack Rudd]

My impression is that playing the person seeded next to you is somewhat more likely under ECF pairings than FIDE ones (they're more likely to be the medians of two consecutive score-groups than one top and one bottom respectively), but I haven't got data to back this up.

[Stewart Reuben]

In the Dutch Open we played sections of 6 RR and then 3-5 (I can't remember) rounds of Swiss. It was before there was a FIDE Rating List, but I didn't notice glaring discrepancies.
RR is much better than Swiss for certain purposes. But useless for apparently high prizes.
Another system has been tried in Birmingham long before Alex was born. Say 64 players 1v64, 63v2,...33v32. The system was used for every score group for every round. I liked it. All the top players won prizes. Err...I wasn't number 64. It is used for the World Cup, but - you take up the position of your opponent if you are lower ranked. Thus 64 beats 1, he becomes number 1. That could also be used for a Swiss, but only one Hastings ever did so as far as I know.

[Alex Holowczak]

In the Dutch Open we played sections of 6 RR and then 3-5 (I can't remember) rounds of Swiss. It was before there was a FIDE Rating List, but I didn't notice glaring discrepancies.

This is a very good way of decelerating the pairings. E.g. 16 players, 6 rounds. 4 groups of 4 (determined by seed). After 3 rounds you get 3/2/1/0 in each group (say). So you get 8 on each scoregroup. By contrast, after 3 rounds in a Swiss with these numbers, you get 4 on 3 and 0, and 12 on 2 and 1. So you've effectively slowed the tournament down by 1 round.

Another system has been tried in Birmingham long before Alex was born. Say 64 players 1v64, 63v2,...33v32. The system was used for every score group for every round. I liked it.

This was also used at my secondary school (in Birmingham) in about 2005 when we did the UK Chess Challenge, because the guy doing the pairings didn't know the rules, and he thought that was how you were supposed to do it.

It is used for the World Cup, but - you take up the position of your opponent if you are lower ranked. Thus 64 beats 1, he becomes number 1. That could also be used for a Swiss, but only one Hastings ever did so as far as I know.

The system you describe is very common in knockout tournaments for every sport. The Chess World Cup seeds all 128 players, though, which is fairly unique. For example, the recent (PDC) World Darts Championship. There were 32 players seeded by world ranking out of 72 players. There were 8 preliminary games. So the First Round had 32 matches. The bracket was drawn up such that the Second Round pairings would, if won by the seeds, be 1v32, 16v17, 8v25, 9v24, 4v29, 13v20, 5v28, 12v21, 2v31, 15v18, 7v26, 10v23, 3v30, 14v19, 6v27, 11v22. The Third Round draw would be Winner Match 1 v Winner Match 2, Winner Match 3 v Winner Match 4 and so on. (The First Round pairings are a seed against a randomly drawn other player who qualified.) This describes exactly what you've just explained, but it might be a better way of visualising the effect of what you wrote! Here is said bracket.

In snooker, the World Championship runs to the same system as the darts, but there are only 16 seeds in 32 players. The same format is used in the Masters, which has 8 seeds in 16 players.

In tennis, the Grand Slams use a modified system of this. For example, the Singles competitions have 32 seeds out of 128. The seeds are subdivided into 5 groups:
Group 1: 1-2
Group 2: 3-4
Group 3: 5-8
Group 4: 9-16
Group 5: 17-32

In Group 1, the #1 seed goes in the top half, and #2 in the bottom half.
This leaves two quarters in the draw without seeds. Each quarter is assigned one seed from Group 2 drawn by lot.
This leaves four eighths in the draw without seeds. Each eighth is assigned one seed from Group 3 drawn by lot.

And so on, until all five groups have been added to the draw. No seeds can meet until Round 3, the last 32. The remaining 96 players are added to the draw by lot.

In fact, one of the few knockout tournaments in sport that doesn't use a seeding system of some kind is the FA Cup. Even then, the Premiership and Championship teams are "seeded", because they get byes into Round 3.

[Stewart Reuben]

Since all systems based at all on ratings must show inevitable bias, a lottery Swiss as used everywhere in England until 1965 and not completely replaced until 1984 will be fairer. But in one tournament it will throw up glaring anomalies.If you repeated 100 times, it would produce truer results.
The seeded System is much more efficient at finding the 'correct' winner.

[E Michael White]

Another system has been tried in Birmingham long before Alex was born. Say 64 players 1v64, 63v2,...33v32. The system was used for every score group for every round. I liked it.

That’s very odd. Back here http://www.ecforum.org.uk/viewtopic.php ... 04&#p64404

you stated that you thought such a system was totally flawed. Perhaps you meant to post in the next thread.

[Roger de Coverly]

Since all systems based at all on ratings must show inevitable bias, a lottery Swiss as used everywhere in England until 1965 and not completely replaced until 1984 will be fairer. But in one tournament it will throw up glaring anomalies.If you repeated 100 times, it would produce truer results.
The seeded System is much more efficient at finding the 'correct' winner.

At one stage, the British Championship combined the two for its early rounds. You would seed the top half and randomly draw their opponents. Such a method should be straightforward enough to program and the use of a key and pseudo-random numbers would make it so the pairings could be replicated by a third party if needed. The historic reluctance of British arbiters to having anything to do with computers would have made that a non-starter.

The real problem is having players "infinitely" better than others in the same tournament. If you want to avoid pairing them, then you need rules which either directly or indirectly avoid this.

[Stewart Reuben]

E Michael White
>Another system has been tried in Birmingham long before Alex was born. Say 64 players 1v64, 63v2,...33v32. The system was used for every score group for every round. I liked it.
That’s very odd. Back here viewtopic.php?f=25&t=3098&p=64404&#p64404
you stated that you thought such a system was totally flawed. Perhaps you meant to post in the next thread.<

Tut. E M W you did not include all I said. After I liked it. All the top players won prizes. Err...I wasn't number 64.
This was what in extremely technical terms is called a joke. Of course I was one of the players grossly favoured by the system.

[E Michael White]

This was what in extremely technical terms is called a joke

Exactly! I meant the joke was odd and probably a reference to Paul McCartney's hit.

[PeterTurland]

Who needs jokes when we have a tribe running our planet?

[Paolo Casaschi]

Since all systems based at all on ratings must show inevitable bias, a lottery Swiss as used everywhere in England until 1965 and not completely replaced until 1984 will be fairer. But in one tournament it will throw up glaring anomalies.If you repeated 100 times, it would produce truer results.
The seeded System is much more efficient at finding the 'correct' winner.

A possible idea could be to use a "tournament Elo" for seeding, starting from the Elo FIDE and updating at every round (possibly with a more dynamic K factor), similarly to the Keyser system (that however uses the "tournament Elo" for scoring and starts from equal numbers).
The advantage would be that the better a player is performing the higher his seeding throughout the tournament.

[Stewart Reuben]

I have thought of the system Peter suggests. Perhaps using K=70 where the rating would turn over in 10-11 games.
It could be computer-simulated. I suspect the effect would be small except for those players who started well and fell away, or those who started badly and made substantial progress later.
I am not certain why David Welch dislikes Dubov. Possibly because it only tries to balance of the strength of the opposition for half the players in any round.
After 2-4 rounds and subsequent ones the players due the white pieces are ranked in descending rating order. The players due black are ranked in ascending Rating Average order. Then paired together. It relies heavily on accurate FIDE Ratings.