Tournament Structures

Discuss anything you like about chess related matters in this forum.
User avatar
Chris Goodall
Posts: 331
Joined: Sun Oct 10, 2010 6:40 pm
Contact:

Re: Tournament Structures

Post by Chris Goodall » Tue Apr 10, 2018 10:33 pm

Michael Farthing wrote:
Tue Apr 10, 2018 9:36 pm
Gentle men,

Though with considerable time and effort I would probably be able to understand this thread, it is currently totally opaque to me.

However, the manner of its conduct is a model of scholarship, decorum and good manners and I heartily congratulate you on this grand achievement and pronounce it Thread of the Year.

Thus, the at first sight erroneous presence of a space in the first line of my post is not a fault but of deliberate intent.
Why thank you! I did ask myself whether the Swiss pairing rules could be understood as an attack on me personally, Northerners, ordinary club players, marginalised identities or Western liberal democracy; but it was too big a stretch even for me :wink:
Last edited by Chris Goodall on Tue Apr 10, 2018 10:40 pm, edited 1 time in total.
Chris Goodall, formerly known as Chris Wardle. ECF Grader for the ancient kingdom of Bernicia (or Northumberland and Durham, if you prefer).
Newcastle is not in Scotland!

NickFaulks
Posts: 4913
Joined: Sat Jan 02, 2010 1:28 pm

Re: Tournament Structures

Post by NickFaulks » Tue Apr 10, 2018 10:34 pm

Chris Goodall wrote:
Tue Apr 10, 2018 8:47 pm
This could be to the benefit of a player who needed to make 2300 for 1 day to get a title.
I can assure you that there are other ways of achieving that which are of far greater concern.

Alex Holowczak
Posts: 8772
Joined: Sat May 30, 2009 5:18 pm
Location: Oldbury, Worcestershire
Contact:

Re: Tournament Structures

Post by Alex Holowczak » Tue Apr 10, 2018 10:41 pm

One of the fundamental assumptions of a Swiss might be flawed - using a seed throughout the tournament.

I've done some analysis of the Chess World Cup results, and the UK Snooker Championship results; both of which use the same pairing system for their knockout tournament. If you plot a route difficulty graph, based on an initial seeding and expected results, then you end up with some logic in the high seeds (i.e. the easiest route is for the player seeded 1), but the gradient of the graph is often negative in places as you work your way through the seeds.

I think I've concluded that seeding 128 players in a 128-player knockout doesn't work from a "fairness" perspective, with the possible caveat that it works if you re-seed after each round; so the highest remaining seed plays the lowest remaining seed in each round, a bit like the NFL playoffs. Both formats are fine if you want media coverage; in the case of snooker, some amateur beating the reigning UK Champion/World Champion/high-ranked player in Round 1 once every few years might be worth the several years of drubbings they get. The same would apply in chess, I guess. 32 would work better, although you might still want to seed 33-64 to keep them apart from 1-32 in Round 1, but draw them randomly rather than using the serpentine method. There's no reason to then seed 65-128 at all.

This got me thinking, maybe the same applies in Swiss tournaments? Quite aside from the pairing principle of highest-seed versus lowest-seed in a scoregroup, rather than 1 v n+1 in a scoregroup with 2n players, maybe there's an over-dependence on the initial seeding? Maybe grouping seeds together might be better; we sort of do that already when we think about "top half against bottom half", but maybe there's an extension of that principle (e.g. keeping people in the same quarter apart) for big tournaments that might apply?

Roger de Coverly
Posts: 17943
Joined: Tue Apr 15, 2008 2:51 pm

Re: Tournament Structures

Post by Roger de Coverly » Tue Apr 10, 2018 10:58 pm

Alex Holowczak wrote:
Tue Apr 10, 2018 10:41 pm
maybe there's an over-dependence on the initial seeding?
I hope I'm remembering this correctly, but didn't larger FIDE rated opens in continental Europe in the 1990s use Buchholz as the ranking order in later rounds? It gave ample opportunity for bouncing ball effects as five games of my initial rating were against 1*GM, 2*IM, 2*FM. I got two draws against them, which was enough for a partial and got 4/4 or 3.5/4 when I faced the unrated 1800s or below.

https://en.wikipedia.org/wiki/Buchholz_system

Outside the UK, computer based pairing systems were becoming established by the early 1990s.

User avatar
Chris Goodall
Posts: 331
Joined: Sun Oct 10, 2010 6:40 pm
Contact:

Re: Tournament Structures

Post by Chris Goodall » Tue Apr 10, 2018 11:02 pm

Alex Holowczak wrote:
Tue Apr 10, 2018 10:41 pm
One of the fundamental assumptions of a Swiss might be flawed - using a seed throughout the tournament.

I've done some analysis of the Chess World Cup results, and the UK Snooker Championship results; both of which use the same pairing system for their knockout tournament. If you plot a route difficulty graph, based on an initial seeding and expected results, then you end up with some logic in the high seeds (i.e. the easiest route is for the player seeded 1), but the gradient of the graph is often negative in places as you work your way through the seeds.

I think I've concluded that seeding 128 players in a 128-player knockout doesn't work from a "fairness" perspective, with the possible caveat that it works if you re-seed after each round; so the highest remaining seed plays the lowest remaining seed in each round, a bit like the NFL playoffs. Both formats are fine if you want media coverage; in the case of snooker, some amateur beating the reigning UK Champion/World Champion/high-ranked player in Round 1 once every few years might be worth the several years of drubbings they get. The same would apply in chess, I guess. 32 would work better, although you might still want to seed 33-64 to keep them apart from 1-32 in Round 1, but draw them randomly rather than using the serpentine method. There's no reason to then seed 65-128 at all.
Re-seeding vastly, vastly favours the top teams/players. If you were trying to manipulate your tournament so that the top seed won, re-seeding combined with top vs. bottom pairings would be literally the best possible way to do it. In every round after the first, the top seed is guaranteed an easier pairing if you re-seed than if you don't.

If you want your tournament to identify the best team/player over the course of the tournament, rather than the best team/player over the rating period preceding the tournament, then you have to honour the scores. You have to take the 127th seed's brilliant victory over the 2nd seed as evidence that the 127th seed is playing at least as well as the 2nd seed this week, and should therefore be considered the 2nd seed for the rest of the tournament.
Chris Goodall, formerly known as Chris Wardle. ECF Grader for the ancient kingdom of Bernicia (or Northumberland and Durham, if you prefer).
Newcastle is not in Scotland!

Kevin Thurlow
Posts: 2838
Joined: Wed Apr 30, 2008 12:28 pm

Re: Tournament Structures

Post by Kevin Thurlow » Wed Apr 11, 2018 8:34 am

"I hope I'm remembering this correctly, but didn't larger FIDE rated opens in continental Europe in the 1990s use Buchholz as the ranking order in later rounds?"

That does sound familiar. They use Buchholz for everything... There is also the Dubov system, https://www.fide.com/fide/handbook.html ... ew=article , summarizing, rankings are ordered by average rating of opponent, after the "number of points". This produces different problems from a normal Swiss! All systems are good and bad.

On Alex's KO systems, Wimbledon tennis used to have top 32 seeded, so that 1 v 32, 2 v 31 etc happened, but they did change to 1 - 16 played anyone from 17 - 32, then 1 - 8, played anyone from 9 - 16 (but keeping their position in the framework, if the "wrong" player won) etc. And they started from a position of 1 - 32 were put in place then all others were fitted round them. So number 32 could play 33 in one pairing, and 127 could play 128 in another. I think World Tennis partly did this when the top 4 were fairly static, so the same semi-finals happened tournament after tournament. Some poor guy played Laver every semi-final.

User avatar
Chris Goodall
Posts: 331
Joined: Sun Oct 10, 2010 6:40 pm
Contact:

Re: Tournament Structures

Post by Chris Goodall » Wed Apr 11, 2018 9:07 am

Alex Holowczak wrote:
Tue Apr 10, 2018 10:41 pm
I think I've concluded that seeding 128 players in a 128-player knockout doesn't work from a "fairness" perspective, with the possible caveat that it works if you re-seed after each round; so the highest remaining seed plays the lowest remaining seed in each round, a bit like the NFL playoffs. Both formats are fine if you want media coverage; in the case of snooker, some amateur beating the reigning UK Champion/World Champion/high-ranked player in Round 1 once every few years might be worth the several years of drubbings they get. The same would apply in chess, I guess. 32 would work better, although you might still want to seed 33-64 to keep them apart from 1-32 in Round 1, but draw them randomly rather than using the serpentine method. There's no reason to then seed 65-128 at all.
Thinking through Alex's comments a bit more, perhaps the Wharton article is measuring the wrong thing when it asks how likely the top k players are to finish in the top k positions.

If you want to find out which player is the best at chess, then the fairest possible tournament is no tournament at all. The rating system, which takes 30 or more games into account, is already a better indication of chess strength than anything you can do in 5 games over the course of a weekend. So the fairest tournament would be one where you gather the players in a hall on the Friday night, give the prizes to the players with the highest grades, and go home.

From that perspective the point of a tournament is to deliberately engineer the possibility that a player who isn't the strongest will win.

I think part of the confusion has been that the Wharton article assumes there's an underlying distribution of strengths that is unknown, and that we're trying to discover through a sampling exercise. Whereas in chess we know the underlying distribution of strengths, more or less, and what we're sampling is performance over the course of the tournament. It's the old sample vs. population business.

So while random pairings, according to the article, may produce information that's as good as or better than a Swiss at discovering the underlying distribution, what we need to ask is whether the Swiss system performs better than random pairings in generating the most meaningful Buchholz scores possible.
Chris Goodall, formerly known as Chris Wardle. ECF Grader for the ancient kingdom of Bernicia (or Northumberland and Durham, if you prefer).
Newcastle is not in Scotland!

User avatar
Chris Goodall
Posts: 331
Joined: Sun Oct 10, 2010 6:40 pm
Contact:

Re: Tournament Structures

Post by Chris Goodall » Wed Apr 11, 2018 9:23 am

Chris Goodall wrote:
Tue Apr 10, 2018 4:14 pm
I hadn't looked at the July 2017 FIDE Swiss pairing rules. Wow, that's a lot of rules. I guess that makes all previous research into chess pairings irrelevant, then. Given the stipulation that there is now only one valid pairing and any human or computer applying the rules must arrive at that one.

EDIT - okay, so we can't compare random pairings with "the Swiss system as it's actually implemented in chess tournaments", but I still think it can be implemented better than the Wharton article managed.
Paul Dargan wrote:
Mon Apr 09, 2018 12:35 am
I have a feeling that GM Smerdon (a serious aceademic economist, when not playing/writing about dodgy Bg4 Scandi lines) has written about the efficiency (or otherwise) of swiss systems... perhaps on chess.com?

Paul
I can find various writings by GM Smerdon on tie-break systems (Fixing Flaws and Stopping Draws). Is that what you were thinking of?
Having now read through and almost understood Smerdon's proposal, assuming the transitivity of results (A > B > C therefore A > C) could be a useful way to wring some extra information out of a set of results, but the big issue I have with it is the assumption that draws = uncombative chess. If you're in a worse position, it takes more combative chess to come away with a draw than a defeat.
Chris Goodall, formerly known as Chris Wardle. ECF Grader for the ancient kingdom of Bernicia (or Northumberland and Durham, if you prefer).
Newcastle is not in Scotland!

User avatar
Michael Farthing
Posts: 1780
Joined: Fri Apr 04, 2014 1:28 pm
Location: Morecambe, Europe

Re: Tournament Structures

Post by Michael Farthing » Wed Apr 11, 2018 10:05 am

Ah, something I fee drawn to comment on.

There are a number of players, consistently rated below me, to whom I almost invariably lose. Might be psychological, might be a playing style difference, probably a bit of both. But if the transitivity assumption is false then the 'correct' result of a tournament is likely to be different from the rating list result (so we would have to play the tournaments rather than just give out the prizes on Friday night). Has it been invetigated properly?

Alex Holowczak
Posts: 8772
Joined: Sat May 30, 2009 5:18 pm
Location: Oldbury, Worcestershire
Contact:

Re: Tournament Structures

Post by Alex Holowczak » Wed Apr 11, 2018 10:20 am

Chris Goodall wrote:
Tue Apr 10, 2018 11:02 pm
Re-seeding vastly, vastly favours the top teams/players. If you were trying to manipulate your tournament so that the top seed won, re-seeding combined with top vs. bottom pairings would be literally the best possible way to do it. In every round after the first, the top seed is guaranteed an easier pairing if you re-seed than if you don't.
This isn't quite right. If all the expected winners win, then you get the same pairings whether you re-seed or not.

You're right that it would favour the best teams/players though - that's precisely the point of re-seeding. You might get one player that does well in the event because all the top seeds have been cleared out of the way due to upset losses in other matches. To take one random example, Joe Swail in the 2015 UK was seeded 57, but reached the last 32. He overachieved. He beat the 72 seed 6-0 in Round 1, but didn't have to play the 7 seed, Ding Junhui in Round 2 - he contrived to lose 6-2 to Adam Duffy, who was a replacement player because there weren't 128 entries. Swail won that 6-5 before losing in the last 32 to the 25 seed, Michael Holt. Had they re-seeded, Duffy would have played the 1 seed, Stuart Bingham in Round 2.
Kevin Thurlow wrote:
Wed Apr 11, 2018 8:34 am
On Alex's KO systems, Wimbledon tennis used to have top 32 seeded, so that 1 v 32, 2 v 31 etc happened, but they did change to 1 - 16 played anyone from 17 - 32, then 1 - 8, played anyone from 9 - 16 (but keeping their position in the framework, if the "wrong" player won) etc. And they started from a position of 1 - 32 were put in place then all others were fitted round them. So number 32 could play 33 in one pairing, and 127 could play 128 in another. I think World Tennis partly did this when the top 4 were fairly static, so the same semi-finals happened tournament after tournament. Some poor guy played Laver every semi-final.
I've just looked into this. The earliest seeding system they had - sort of - was to give the defending champion the number 1 seed, and give him a bye to the Final. Everyone else had to playoff for the right to play him. A bit like current Chess World Championship.

Wimbledon's first proper seeding system was in 1927. They had 8 of them, but they don't seem to have been drawn to result in Quarter Finals of 1 v 8, 4 v 5, 2 v 7 and 3 v 6. In fact, you'd have got 2 v 7, 4 v 6, 1 v 8, 3 v 5. I think the seeds were drawn to keep the 1 and 2 apart until the Final; the 1 to 4 apart until the Semi Finals, and the 1 to 8 apart until the Quarter Finals.

Wimbledon experimented with different numbers of seeds in the early 1950s, jumping from 8 to 16, then down to 10, up to 12 again, then back down to 8. In 1968, they went up to 16, and in 2001, they went up to 32. But they've never strictly seeded them. They've just kept 1-2 apart until the Final, 1-4 apart until the Semis, 1-8 apart until the Quarters, 1-16 apart until the Last 16, and 1-32 apart until the Last 32. They appear to have done this ever since they had a seeding system, since 1927.

This is perhaps a way to improve the Chess World Cup pairing system if "fairness" is the aim, you just seed put use that above principle as at Wimbledon, but extend it for 64 seeds. It makes sense to draw them randomly, given that if you look at the graph between 33 and 64, the seeds in the 60s have an easier route than some seeds in the 50s.

Kevin Thurlow
Posts: 2838
Joined: Wed Apr 30, 2008 12:28 pm

Re: Tournament Structures

Post by Kevin Thurlow » Wed Apr 11, 2018 10:44 am

" I think the seeds were drawn to keep the 1 and 2 apart until the Final; the 1 to 4 apart until the Semi Finals, and the 1 to 8 apart until the Quarter Finals."

So it was accidental when they got 1v8, 2v7 etc! Wimbledon also ignored the world rankings when they felt like it. Everyone else just used the world rankings, but Wimbledon said "Lendl? Can't play on grass, we're not making him no. 1!" Eventually they fell into line...This is not really a factor in chess of course.

"There are a number of players, consistently rated below me, to whom I almost invariably lose. Might be psychological, might be a playing style difference, probably a bit of both."

I think that applies to a lot of people. There are one or two annoying individuals to seem to play better against me than against other people. Still, there was a Swiss guy at least 200 higher than me that I beat 3 times in a row increasingly quickly. That's why we play chess instead of just reading the grading list.

Alex Holowczak
Posts: 8772
Joined: Sat May 30, 2009 5:18 pm
Location: Oldbury, Worcestershire
Contact:

Re: Tournament Structures

Post by Alex Holowczak » Wed Apr 11, 2018 11:10 am

Kevin Thurlow wrote:
Wed Apr 11, 2018 10:44 am
Wimbledon also ignored the world rankings when they felt like it. Everyone else just used the world rankings, but Wimbledon said "Lendl? Can't play on grass, we're not making him no. 1!" Eventually they fell into line...This is not really a factor in chess of course.
Wimbledon still does that, to the best of my knowledge. Its seeds are based on world ranking and performance in grass court tournaments.

User avatar
Chris Goodall
Posts: 331
Joined: Sun Oct 10, 2010 6:40 pm
Contact:

Re: Tournament Structures

Post by Chris Goodall » Wed Apr 11, 2018 11:23 am

Alex Holowczak wrote:
Wed Apr 11, 2018 10:20 am
Chris Goodall wrote:
Tue Apr 10, 2018 11:02 pm
Re-seeding vastly, vastly favours the top teams/players. If you were trying to manipulate your tournament so that the top seed won, re-seeding combined with top vs. bottom pairings would be literally the best possible way to do it. In every round after the first, the top seed is guaranteed an easier pairing if you re-seed than if you don't.
This isn't quite right. If all the expected winners win, then you get the same pairings whether you re-seed or not.
Counter-nitpick: if all the expected winners win, then you can't call what you're doing re-seeding.
Alex Holowczak wrote:
Wed Apr 11, 2018 10:20 am
You're right that it would favour the best teams/players though - that's precisely the point of re-seeding.
So why is that a desirable thing to do? The fact that you're holding a chess tournament and not a darts tournament favours the best chess players. They're already the most likely to win, because they're the best at the skill that you're measuring. The best possible tournament is one in which their greater likelihood of winning is exactly in proportion to their greater skill, no less and no more. Just as the ideal tournament shouldn't compress differences in strength, neither should it stretch them.
Last edited by Chris Goodall on Wed Apr 11, 2018 11:43 am, edited 2 times in total.
Chris Goodall, formerly known as Chris Wardle. ECF Grader for the ancient kingdom of Bernicia (or Northumberland and Durham, if you prefer).
Newcastle is not in Scotland!

Mick Norris
Posts: 7376
Joined: Tue Apr 17, 2007 10:12 am
Location: Bolton, Greater Manchester
Contact:

Re: Tournament Structures

Post by Mick Norris » Wed Apr 11, 2018 11:25 am

Alex Holowczak wrote:
Wed Apr 11, 2018 11:10 am
Kevin Thurlow wrote:
Wed Apr 11, 2018 10:44 am
Wimbledon also ignored the world rankings when they felt like it. Everyone else just used the world rankings, but Wimbledon said "Lendl? Can't play on grass, we're not making him no. 1!" Eventually they fell into line...This is not really a factor in chess of course.
Wimbledon still does that, to the best of my knowledge. Its seeds are based on world ranking and performance in grass court tournaments.
I think for the men, but not for the women? I recollect the women's seedings were on world rankings only, but the men also took into account grass court performance
Any postings on here represent my personal views and should not be taken as representative of the Manchester Chess Federation www.manchesterchess.co.uk

Alex Holowczak
Posts: 8772
Joined: Sat May 30, 2009 5:18 pm
Location: Oldbury, Worcestershire
Contact:

Re: Tournament Structures

Post by Alex Holowczak » Wed Apr 11, 2018 1:28 pm

Chris Goodall wrote:
Wed Apr 11, 2018 11:23 am
Alex Holowczak wrote:
Wed Apr 11, 2018 10:20 am
You're right that it would favour the best teams/players though - that's precisely the point of re-seeding.
So why is that a desirable thing to do? The fact that you're holding a chess tournament and not a darts tournament favours the best chess players. They're already the most likely to win, because they're the best at the skill that you're measuring. The best possible tournament is one in which their greater likelihood of winning is exactly in proportion to their greater skill, no less and no more. Just as the ideal tournament shouldn't compress differences in strength, neither should it stretch them.
In which case, you might as well just pair any tournament completely randomly, and not seed anything.

You're focusing on the impact of the 1, rather than the field. I think it's reasonable that in a seeded knockout tournament, the 1 gets an easier route than the 128. I mean, the 1 never has to play the 1, for starters, which makes it easier than every other seed. If you seed all 128 players, and you would expect the 1 to have the easiest route and 128 to have the hardest route, you might expect a nice straight line connecting the two together. But you don't, you get a graph with curves and changes of direction. Why should seed 90 have a harder route than seed 120? I don't think that's a feature of what one might describe as "the ideal tournament".

The NFL playoffs might justify it by saying that it's a reward for being the 1 that you get the easiest route to the playoffs; a way of ensuring teams continue to play seriously at the end of the regular season when they've already qualified for the playoffs, because the number 1 seed is worth competing for. That logic might also apply to the FIDE rating list, to make getting the number 1 seed something worth competing for. It doesn't work in the UK Snooker Championship, where the 1 is the defending champion rather than the player ranked 1. The logic might reasonably apply in chess too.

Post Reply