FIDE rated events 2011

[Roger de Coverly]

Apparently, the pairings are done for the next round of a tournament by accessing a central computer at the FFE (French equivalent of the ECF), which sounds very impressive.

I'd suspect in practice, pairings are done locally on the arbiter's computer using FFE software previously downloaded. You need to be wary of the French approach to acceleration though. Top quartile players get a dummy point for the first couple of rounds, so a first round draw can result in a pairing against another top quartile player who also drew.

It's feasible to do UK pairings by computer - I believe e2e4 use them. Many British arbiters believe the card approach generates "better" pairings than the computer programs.

[Alex Holowczak]

It's feasible to do UK pairings by computer - I believe e2e4 use them. Many British arbiters believe the card approach generates "better" pairings than the computer programs.

e2e4 use the International pairing system. Sean can perhaps explain why; my guess is that that's what Swiss Master does and/or they use it because it's a FIDE-rated event.

Many British arbiters are right if the software you're comparing it to is Tournament Director, which sometimes gets floats wrong. Swiss Master seems to have the FIDE pairing algorithm nailed though.

[Sean Hewitt]

It's feasible to do UK pairings by computer - I believe e2e4 use them. Many British arbiters believe the card approach generates "better" pairings than the computer programs.

e2e4 use the International pairing system. Sean can perhaps explain why; my guess is that that's what Swiss Master does and/or they use it because it's a FIDE-rated event.

Many British arbiters are right if the software you're comparing it to is Tournament Director, which sometimes gets floats wrong. Swiss Master seems to have the FIDE pairing algorithm nailed though.

We do use computer pairings for a number of reasons

- They don't make basic mistakes (such as pairing players twice, or illegal colours etc)
- SwissMaster appears to get floats etc spot on
- The computer generates the website data and skeleton pgn files without any extra work
- The computer generates a FIDE rating report too

All in all, the computer makes life easier.

We pair to FIDE pairing rules rather than ECF rules for two reasons. Firstly, it means that our players from abroad understand the parings but more importantly, I (and the rest of the world!) believe that FIDE pairing rules are more sensible than ECF pairing rules, particularly regarding colour sequences and choice of player to float.

[Richard Bates]

We pair to FIDE pairing rules rather than ECF rules for two reasons. Firstly, it means that our players from abroad understand the parings but more importantly, I (and the rest of the world!) believe that FIDE pairing rules are more sensible than ECF pairing rules, particularly regarding colour sequences and choice of player to float.

Rest of the World are wrong! Middle floats is much better.

[IM Jack Rudd]

I (and the rest of the world!) believe that FIDE pairing rules are more sensible than ECF pairing rules, particularly regarding colour sequences and choice of player to float.

It depends what you want out of your pairing system. I, for example, don't want a system in which the weakest player in the tournament will be given a full-point bye in the first round: the result of such a system is to give such a player an inappropriately strong set of opponents for the first few rounds of the tournament.

(That is the only part of the system where I actively dislike what FIDE's floating rules do. Anywhere else in the tournament, I don't have strong enough feelings either way to actively advocate anything; I'd like, one of these days, to run some sort of Monte Carlo simulation to see what floating rules produce the best overall results.)

[Richard Bates]

I (and the rest of the world!) believe that FIDE pairing rules are more sensible than ECF pairing rules, particularly regarding colour sequences and choice of player to float.

It depends what you want out of your pairing system. I, for example, don't want a system in which the weakest player in the tournament will be given a full-point bye in the first round: the result of such a system is to give such a player an inappropriately strong set of opponents for the first few rounds of the tournament.

(That is the only part of the system where I actively dislike what FIDE's floating rules do. Anywhere else in the tournament, I don't have strong enough feelings either way to actively advocate anything; I'd like, one of these days, to run some sort of Monte Carlo simulation to see what floating rules produce the best overall results.)

This little discussion probably needs transporting elsewhere, but anyway.

I find the ECF rules on floats entire logical and consistent with the general Swiss Pairing principles of top half vs bottom half, and the FIDE rules otherwise, and consequently providing the least unfairness to those affected.

Firstly the point about the bye makes this most obvious - it makes no logical sense for the two lowest seeds in the score group (bottom seed and the bye) to be paired against each other. To pair in this way is only really justified if you see the bye as a "penalty" which the lowest seed should suffer for being the weakest player (some tournaments give half point byes at the bottom which clearly is an attempt to mitigate this - although this can cause other serious problems especially in team events, where genuine draws are less common - this was demonstrated at the European Club Cup where the bye receipients soon formed their own mini subgroup for pairing purposes resulting in some of them getting far more points than they would have done had there been no bye in the tournament).

Secondly in a "normal" scenario one can think about what different players in a score group can normally expect from the pairings. Except in extreme cases (and/or cases with limited numbers) a player seeded at the top will never expect to play the weakest possible player, and a player at the bottom will never expect to play the strongest possible player. A median player however can expect to play either depending on which side of the top/bottom divide they fall. Consequently i see it as entirely justified, least disruptive and least unfair that the median player should float.

[Roger de Coverly]

Firstly the point about the bye makes this most obvious - it makes no logical sense for the two lowest seeds in the score group (bottom seed and the bye) to be paired against each other.

Agreed on this - Isn't there a fairly obvious hack though? Identify the issue that you have an odd number of players. Identify the median and remove from the draw by giving the player an artificial half point bye. Pair the remaining even number of players. Reinstate the median player with a full point bye or a game against a filler.

Regarding regular floats, I don't see the logical problem with pairing the lowest ranked player on (n) with the highest ranked player on (n-0.5). It's quite likely in practice in a large score group that the player with the lower score will have the higher rating. Byes are different and giving the bye (or lower rated filler) to the median remains logical even after the first round.

[Richard Bates]

Firstly the point about the bye makes this most obvious - it makes no logical sense for the two lowest seeds in the score group (bottom seed and the bye) to be paired against each other.

Agreed on this - Isn't there a fairly obvious hack though? Identify the issue that you have an odd number of players. Identify the median and remove from the draw by giving the player an artificial half point bye. Pair the remaining even number of players. Reinstate the median player with a full point bye or a game against a filler.

Whether this is a viable solution or not isn't really relevant because it's not, to my understanding, FIDE pairing rules - the merits or otherwise of which are being discussed.

Regarding regular floats, I don't see the logical problem with pairing the lowest ranked player on (n) with the highest ranked player on (n-0.5). It's quite likely in practice in a large score group that the player with the lower score will have the higher rating. Byes are different and giving the bye (or lower rated filler) to the median remains logical even after the first round.

Well my argument for this was addressed in my second paragraph - and it's almost the opposite extreme to, but not in contradiction of, the situation where a bye is floating about. With a bye the weakest player gets an unjustifiably 'easy' game, without the weakest player gets an unjustifiably tough game. In principle i don't see why the system should be designed in the event of floats to unduly advantage "stronger" players who drop points and disadvantage "weaker" players who don't (or indeed any player who gets the downfloat).

[Roger de Coverly]

Whether this is a viable solution or not isn't really relevant because it's not, to my understanding, FIDE pairing rules - the merits or otherwise of which are being discussed.

In part this discussion started with trying to strip away all the excuses used by British arbiters for not using computerised pairings, one of which is the dislike of the common practice under FIDE rules of giving the bye to the lowest ranked player. My point being that you can use computerised pairings without having to compromise on the bye.

I don't find later round issues regarding the morality of floats particularly interesting. It's likely that whatever rule is applied is likely to be distorted by not playing the same opponent twice or colour distribution and sequence issues.

[Alex Holowczak]

Another reason - it's not really an excuse - would be avoiding the pairing of relatives/club mates in weekend events. This is an element of pairing that hasn't quite been catered for by computers yet.

That's another matter of taste. Most of the time, I don't think it's a problem worth worrying about; they'll give each other a good game all the same. Some people who play for two clubs always enter using the name of the club they perceive to be strongest, so that they avoid better players.

[Sean Hewitt]

Firstly the point about the bye makes this most obvious - it makes no logical sense for the two lowest seeds in the score group (bottom seed and the bye) to be paired against each other.

I don't understand this point at all. Surely, if there is a bye, he is not paired against anyone?

[IM Jack Rudd]

Firstly the point about the bye makes this most obvious - it makes no logical sense for the two lowest seeds in the score group (bottom seed and the bye) to be paired against each other.

I don't understand this point at all. Surely, if there is a bye, he is not paired against anyone?

Richard is, if I understand him correctly, modelling the bye as a virtual player who is weaker than anybody else in the tournament, and on a lower score than anybody else in the tournament. Both the ECF and the FIDE systems effectively do this when it comes to assigning byes.

[Richard Bates]

Firstly the point about the bye makes this most obvious - it makes no logical sense for the two lowest seeds in the score group (bottom seed and the bye) to be paired against each other.

I don't understand this point at all. Surely, if there is a bye, he is not paired against anyone?

Richard is, if I understand him correctly, modelling the bye as a virtual player who is weaker than anybody else in the tournament, and on a lower score than anybody else in the tournament. Both the ECF and the FIDE systems effectively do this when it comes to assigning byes.

If you go into Megabase player listings and sort by games played, who comes top?

Yes Jack is correct - and it is by thinking of it like this that the ECF system is most clearly justified. When pairing you do one score group at a time. Where there is an uneven number of players you add an extra player (the upfloat) to create an even number. The upfloater playing the middle player is consistent with the upfloater being slotted in as the lowest seed on the score group. If the middle player downfloats it is logical that a middle player should also upfloat. Basically i think the ECF system is more consistent with general Swiss pairing principles.

This isn't of course the only way you could do it, especially if you dislike the "assumption" that the upfloater becomes the bottom seed. A reasonable, in principle, alternative IMO would be to always seek to upfloat the highest player and then pair the whole group conventionally (with the upfloat slotting in wherever his (rating) seeding suggests). Ie. unlike conventionally where I think you generally run the process - 1) pair higher group -> 2)determine downfloater -> 3)seek appropriate upfloater, you would run the process- 1) Determine upfloater (possibly without reference to colours if you like a bit of sacrilege!) -> 2) pair higher group. I suspect that would be a lot more complicated in practice, however.

It could be argued (maybe this is one of the reasons Sean likes it for his tournaments?) that the FIDE system is better for securing outright winners (especially in tournaments with "too many" players for the rounds available), because it apparently shares its underlying logic with accelerated pairing systems ie. it is designed so that strong players on lower scores will "knock out" the weakest players on the higher scores.

Of course i have never paired a round in my life, never properly studied the operation of the systems, so this can all be taken as the random musings of a player who probably doesn't really know what he's talking about!

[Alex McFarlane]

Richard,

You are generally correct in what you say about pairings and for the reasons that you state.

With regard to SwissMaster, I do not believe that it always comes up with the best pairings. Twice in the last year (admittedly with a relatively small number of players) it has handled floats badly. On both occasions I have had a scoregroup with a downfloat into it. The downfloat has resulted in a pairing of the group being impossible so SM has then downfloated from that group into the one below. However by changing the downfloat into the original group no further downfloats would have been necessary.

Currently acceleration by computer program is, to say the least, 'iffy'. A Norwegian player (and arbiter) at Hastings who had queried using acceleration two years ago came up to me this time to say that he had been studying various tournaments and was now convinced of the merits of the system employed in Britain.

I believe that the FIDE pairing system (Dutch system) was simplified to allow for computer programming to take place. The instructions are certainly more computer friendly than human understandable.

[LozCooper]

http://www.escacs.cat/circuit/calendari.htm

Calendar of Catalan circuit tournaments