Suitability of acceleration?

Discuss anything you like about chess related matters in this forum.
Roger de Coverly
Posts: 21301
Joined: Tue Apr 15, 2008 2:51 pm

Re: Suitability of acceleration?

Post by Roger de Coverly » Sat May 15, 2021 11:14 am

Mike Gunn wrote:
Sat May 15, 2021 10:54 am
Pairing Q1 v Q2 and Q3 v Q4 in round 1 removes the worst mis-matches then just carry on as in a normal Swiss.
If all the games go according to the seeding with decisive results, the second round just becomes Q1 v Q3 and Q2 v Q4.

If the ratings are unreliable and thus seedings questionable there's always the possibility of drawing the first round randomly and normal Swiss thereafter.

Ian Thompson
Posts: 3551
Joined: Wed Jul 02, 2008 4:31 pm
Location: Awbridge, Hampshire

Re: Suitability of acceleration?

Post by Ian Thompson » Sat May 15, 2021 11:28 am

Mike Gunn wrote:
Sat May 15, 2021 10:54 am
I have run club tournaments with a similar number of players to the ones you are expecting. Pairing Q1 v Q2 and Q3 v Q4 in round 1 removes the worst mis-matches then just carry on as in a normal Swiss.
Isn't that just going to result in the same pairings as with a normal pairing system, but with the order of rounds 1 and 2 reversed, if results go with seedings?

Code: Select all

Normal pairings: R1: Q1vQ3; Q2vQ4; R2: Q1vQ2; Q3vQ4
Your pairings:   R1: Q1vQ2; Q3vQ4; R2: Q1vQ3; Q2vQ4
If results don't go with seedings in your pairing system you get the same mismatches as you would with a normal pairing system, e.g. a Q1 player loses in round 1 then they play a Q4 player in round 2 with either pairing system, or a Q2 player loses in round 1 then they play a Q4 player in round 2 with either pairing system.

Edit: I see Roger has already made a similar comment.

Roger Lancaster
Posts: 1910
Joined: Tue Mar 17, 2015 2:44 pm

Re: Suitability of acceleration?

Post by Roger Lancaster » Sat May 15, 2021 12:18 pm

Message deleted - replaced below
Last edited by Roger Lancaster on Sat May 15, 2021 12:22 pm, edited 2 times in total.

Roger Lancaster
Posts: 1910
Joined: Tue Mar 17, 2015 2:44 pm

Re: Suitability of acceleration?

Post by Roger Lancaster » Sat May 15, 2021 12:20 pm

Mike Gunn wrote:
Sat May 15, 2021 10:54 am
I have run club tournaments with a similar number of players to the ones you are expecting. Pairing Q1 v Q2 and Q3 v Q4 in round 1 removes the worst mis-matches then just carry on as in a normal Swiss. I don't think anything else is justified (or works in practical terms) with such low numbers.
Maybe Mike was a shade careless in his wording but, after the round 1 pairings he sets out and assuming that those results then go as planned, I would expect round 2 to see {a} the top half of Q1 paired against the bottom half {b} the Q2 players, who had lost in round 1, paired against the Q3 players, who had won in round 1, and {c} the top half of Q4 paired against the bottom half. If the Q2 players perform as expected and win [or at least draw] in round 2, then this results in a maximum of four players reaching 2/2 as opposed to eight with a normal Swiss. Of course, it's fair to add that the actual results typically don't completely follow the hoped-for pattern in which case the acceleration will have been only partially successful.
Last edited by Roger Lancaster on Sat May 15, 2021 12:23 pm, edited 1 time in total.

Mike Gunn
Posts: 1022
Joined: Wed Apr 11, 2007 4:45 pm

Re: Suitability of acceleration?

Post by Mike Gunn » Sat May 15, 2021 12:22 pm

No, I don't think so (because there is usually a small number of upsets/ draws). Q4 players misses out on playing a Q2 players in round 1 and this is an important effect. Anyway, my observation is that you end up with fewer games with large mis-matches between opponents. It is difficult to prove, but I do have several years of data which I could examine ...

Mike Gunn
Posts: 1022
Joined: Wed Apr 11, 2007 4:45 pm

Re: Suitability of acceleration?

Post by Mike Gunn » Sat May 15, 2021 12:29 pm

(In reply to Roger): my actual experience is that you can't make the 2nd round of acceleration work in any meaningful way with less than about 30 players. I started off with the rules from Stewart's book in front of me, intending to do a 2nd round of acceleration and discovered you jujst couldn't make it work/ it produced ridiculous pairings.

Wadih Khoury
Posts: 604
Joined: Sun Jul 12, 2020 8:14 pm

Re: Suitability of acceleration?

Post by Wadih Khoury » Sun May 16, 2021 9:34 am

I ran a few simulations (c. 10, not really statistically meaningful, but just to get an idea), and I now tend to agree with some of the comments above: With the pool of players at hand, all that happens is that acceleration will shift the "ridiculous" pairings from round 1 and 2 to the first 2 rounds post acceleration.

I did notice that acceleration tended to slightly reduce the number of non-meaningful games over the length of the tournament, but it may be due to my sample size (roughly 3 more meaningful games over 45, so likely to be statistical noise)

NickFaulks
Posts: 8453
Joined: Sat Jan 02, 2010 1:28 pm

Re: Suitability of acceleration?

Post by NickFaulks » Sun May 16, 2021 10:30 am

Wadih Khoury wrote:
Sun May 16, 2021 9:34 am
I did notice that acceleration tended to slightly reduce the number of non-meaningful games over the length of the tournament
In current circumstances it is even harder than usual to predict which games will be meaningful. My rule would be not to try to be too clever.
If you want a picture of the future, imagine a QR code stamped on a human face — forever.

Paul McKeown
Posts: 3732
Joined: Thu Apr 12, 2007 3:01 pm
Location: Hayes (Middx)

Re: Suitability of acceleration?

Post by Paul McKeown » Sun May 16, 2021 11:08 am

Angus French wrote:
Fri May 14, 2021 10:59 pm
Paul McKeown wrote:
Fri May 14, 2021 10:32 pm
Wadih Khoury wrote:
Fri May 14, 2021 10:19 am
If I use acceleration, does it have to be Baku (which I believe is the FIDE sanctioned method) in order to be graded by the ECF?
ECF rating doesn't care about your pairing method at all. Do what you want, just ensure that everyone enjoys themselves, and the Laws are upheld.
... and everyone is safe.
Of course. No disagreement with me - all sensible, medically sound and legally required precautions should be in place. I thought it went without saying, but you are, sadly, right. Some disagree with the idea.

Paul McKeown
Posts: 3732
Joined: Thu Apr 12, 2007 3:01 pm
Location: Hayes (Middx)

Re: Suitability of acceleration?

Post by Paul McKeown » Sun May 16, 2021 11:13 am

Acceleration has a primary purpose: to allow the determination of a winner from a Swiss format tournament in which there are insufficient rounds for the number of players. Separation of players with highly divergent ratings is a secondary purpose.

Roger de Coverly
Posts: 21301
Joined: Tue Apr 15, 2008 2:51 pm

Re: Suitability of acceleration?

Post by Roger de Coverly » Sun May 16, 2021 12:58 pm

Paul McKeown wrote:
Sun May 16, 2021 11:13 am
Separation of players with highly divergent ratings is a secondary purpose.
If you wanted to do that as a primary purpose, devising a pairing scheme which modified the usual top down would be an approach. Resorting to the device of having more than one float would be a possibility. So you might have a rule that unless someone was playing well above or possibly well below their initial rating, pairings of more than x rating points apart were barred unless part of a fight for top places in the final rounds. Players unable to get a satisfactory pairing within their scoregroup are floated up or down as necessary.

NickFaulks
Posts: 8453
Joined: Sat Jan 02, 2010 1:28 pm

Re: Suitability of acceleration?

Post by NickFaulks » Sun May 16, 2021 1:29 pm

Paul McKeown wrote:
Fri May 14, 2021 10:32 pm
ECF rating doesn't care about your pairing method at all. Do what you want, just ensure that everyone enjoys themselves, and the Laws are upheld.
At the risk of repeating myself, unless title norms are involved FIDE doesn't care about pairing methods either.
If you want a picture of the future, imagine a QR code stamped on a human face — forever.

Paul McKeown
Posts: 3732
Joined: Thu Apr 12, 2007 3:01 pm
Location: Hayes (Middx)

Re: Suitability of acceleration?

Post by Paul McKeown » Mon May 17, 2021 1:43 am

NickFaulks wrote:
Sun May 16, 2021 1:29 pm
unless title norms are involved FIDE doesn't care about pairing methods either.
Absolutely.

Paul McKeown
Posts: 3732
Joined: Thu Apr 12, 2007 3:01 pm
Location: Hayes (Middx)

Re: Suitability of acceleration?

Post by Paul McKeown » Mon May 17, 2021 1:46 am

Roger de Coverly wrote:
Sun May 16, 2021 12:58 pm
Paul McKeown wrote:
Sun May 16, 2021 11:13 am
Separation of players with highly divergent ratings is a secondary purpose.
If you wanted to do that as a primary purpose, devising a pairing scheme which modified the usual top down would be an approach. Resorting to the device of having more than one float would be a possibility. So you might have a rule that unless someone was playing well above or possibly well below their initial rating, pairings of more than x rating points apart were barred unless part of a fight for top places in the final rounds. Players unable to get a satisfactory pairing within their scoregroup are floated up or down as necessary.
Thank you, Roger. An Intriguing post.

Perhaps you could develop this a little further, with examples even?

It gets to the heart of Wadih's original post, I think, and is perhaps a common requirement not met by current (official) methods.

Roger de Coverly
Posts: 21301
Joined: Tue Apr 15, 2008 2:51 pm

Re: Suitability of acceleration?

Post by Roger de Coverly » Mon May 17, 2021 8:57 am

Paul McKeown wrote:
Mon May 17, 2021 1:46 am
Perhaps you could develop this a little further, with examples even?
In its simplest possible form, just apply it as a marginal distortion to the first round. Rank the players as usual but keep a note of the ratings. Run the usual top v bottom pairing routine. But then check each individual pairing. I'd suspect you would have to start at the top and bottom simultaneously. Set some rating difference of your choice. If the pairing of the top and bottom seeds don't infringe this, continue. Otherwise look down the ranking risk from the top for the top seed and up the ranking list for the bottom seed until an opponent is discovered within the rating difference rule. Mark these pairs as a forced pairing. Repeat the process for the remaining n-2 or n-4 players until all players have pairings within the tolerance.

The same process could be repeated for later rounds, but has to stop at some stage to avoid distortion of the overall results and might result in no complete pairing being possible. At the start, it may be necessary to allocate pairing ratings to players at the top and bottom to bring them within range of a selection of opponents.