Swiss Pairing Question

[Joey Stewart]

There is a slightly unusual tournament situation in the current gloucester championships, which I was wondering if anyone familiar with swiss pairings knew how the pairing is likely to be resolved:

Joint first - two players on 3/3 both due to play as white

Joint second - two players on 2/3 both due to play black

What would take priority in this situation - if it is points scored then is there a way to determine which player gets the double black, or are the second placed players upfloated to balance the colours?

[IM Jack Rudd]

Points scored takes priority.

When deciding who gets white with two players both due to play the same colour, your order of priority goes as follows:

(a) The player whose colour preference is stronger gets his due colour.
(b) If both players have equally strong colour preference, the player on the higher score gets his due colour.
(c) If both players have equally strong colour preference and are on the same score, the player with the higher rating gets his due colour.

[Stewart Reuben]

In case you don't understand Jack's perfectly correct statement.
WBB
BWB
BBW. The first gets the W in round 4, failing that the second, failing that the higher graded or rated.

I have always felt the Swiss is a bit of a swizz. It is normal for more players on high scores to have had more whites than blacks. Many tournaments have an odd number of rounds. Then 3W and 2B, or 2B and 3W is common. If it is 6 rounds, 4W and 2B is not all that uncommon.

Chess really should be played as a double game tournament against the same opponent. These are almost unknown except at th highest levels.

[Niall Doran]

Another Swiss pairing question.

Imagine a seven round Swiss Open, with players from 1000 all the way up to 2600, with 80 players. In the 1st round, the n°1 seed plays n°41 and all the way down to n°40 plays n°80 i.e. no accelerated pairings.

Over an infinite number of tournaments run to these specifications, will n°40 do a lot better than n°41 because of an easier start?

[IM Jack Rudd]

Another Swiss pairing question.

Imagine a seven round Swiss Open, with players from 1000 all the way up to 2600, with 80 players. In the 1st round, the n°1 seed plays n°41 and all the way down to n°40 plays n°80 i.e. no accelerated pairings.

Over an infinite number of tournaments run to these specifications, will n°40 do a lot better than n°41 because of an easier start?

No idea. The only practical way to answer that question will be with a Monte Carlo method.

[Stewart Reuben]

You can computer simulate perhaps 1000 such tournaments and see whether Number 40 does better than 41.
Since it is an odd number of rounds, it may depend on whether they get White or Black in round 1. Both have rather wasted their round 1 White. 40 is 20 rating points higher than 41, so he should get a slighly higher score. It would be better if those two players had the sme rating, with 40 coming first alphabetically. i.e. both 1790.
To make the simulation more meaningful, you draw up an 80 player RR, one with No.1 having white and the other with him having black. You use ChessBase to get the actual results between players of the different ratings. That is less than perfect. In real life, if a player starts well, he tends to continue to do well. You need a random number generator, even for round 1 whre this is a difference of 800 rating points.

But Niall, why ask the question? You don't seriously think a standard seeded Swiss is perfectly fair? Both 40 and 41 will tend to bob up and down.

[MartinCarpenter]

They're actively terrible at sorting out the middle of the field aren't they?

Too many physcological issues for sims to work brilliantly I'd think.

[IM Jack Rudd]

Too many physcological issues for sims to work brilliantly I'd think.

That's why you run the simulation a million times or whatever. Psychological issues will cause noise, and you correct for noise by running huge numbers of sims.

[Niall Doran]

You can computer simulate perhaps 1000 such tournaments and see whether Number 40 does better than 41.
Since it is an odd number of rounds, it may depend on whether they get White or Black in round 1. Both have rather wasted their round 1 White. 40 is 20 rating points higher than 41, so he should get a slighly higher score. It would be better if those two players had the sme rating, with 40 coming first alphabetically. i.e. both 1790.
To make the simulation more meaningful, you draw up an 80 player RR, one with No.1 having white and the other with him having black. You use ChessBase to get the actual results between players of the different ratings. That is less than perfect. In real life, if a player starts well, he tends to continue to do well. You need a random number generator, even for round 1 whre this is a difference of 800 rating points.

But Niall, why ask the question? You don't seriously think a standard seeded Swiss is perfectly fair? Both 40 and 41 will tend to bob up and down.

[Niall Doran]

But Niall, why ask the question? You don't seriously think a standard seeded Swiss is perfectly fair? Both 40 and 41 will tend to bob up and down.

The reason is that I tend to be around n°45 (if we take my example) and was wondering whether the guys a few points stronger than me would have an advantage, and whether this should motivate me to work harder and gain those points!

For another example, every year for the past two years I've played in a five-round, under-2000 tournament over three days. Being in the lower half, I always end up having a tough game on the Friday night, finishing very late (midnight or after) and then with a 9 a.m. start the next day. So here I fantasise about being in the top half and getting a lower-rated opponent, finishing with a win not too late, and up bright and fresh the next day!

To be honest, I realise the Swiss isn't great, having bobbed up and down quite a lot myself in the past.

[Joey Stewart]

I would rather have a swiss then have, for example, a lower middle table player win a tournament through freak luck of the draw seeing him avoid all the top players - everybody would be unhappy in that case, so at least swiss works for the majority of players.

It also makes a huge assumption that players ratings accurately reflect their strength and, as we all know, fide ratings do anything but that....

[Matthew Turner]

Another Swiss pairing question.

Imagine a seven round Swiss Open, with players from 1000 all the way up to 2600, with 80 players. In the 1st round, the n°1 seed plays n°41 and all the way down to n°40 plays n°80 i.e. no accelerated pairings.

Over an infinite number of tournaments run to these specifications, will n°40 do a lot better than n°41 because of an easier start?

In the example that you quote then no. 40 will do significantly better than no. 41 (in terms of points scored). However, I am not sure whether the easier start has a significant effect (I suspect the impact will be very limited). The key factor is the pairing in the last round. In your example it is relative likely that no.s 40 and 41 will have 3/6 and in this situation no. 40 is significantly more likely to be in the top half and get an easier pairing.

[Stewart Reuben]

AS Jonathan Mestel said (relying on my imperfect memory): There is nothing wrong with a Swiss system, What is wrong is assuming the raw score gives an adequate assessment of achievemnt.
You can use TPR instead, That way, somebody with a lower score may get a better TPR because he has met stronger opposition. WE have rating prizes buuilt around this principle. They are usually rating impovement prizes.
It is also possible to iterate.
Takes the players' ratings.
Determine their TPR in the tournament.
Replace their rating by their TPR.
Recalculate the TPR from this data.
You can keep on doing this until the numbers stay the same.
The one time I did this for one British, it made no difference.

Niall, it is up to you to decide whether it is worthwhile working hard to become a bit stronger. Achieving that must be its own reward. In my own experience, and as an organiser, I have worked on the assumption that becoming a better player makes playing any game more enjoyable.
I once wrote an article in the BCM about improving your results without becoming a better player. e.g. a 10 minute walk before the game.

[Niall Doran]

Yet another pairing question.

Assuming the goal of the organiser is for players to have as many games where they play opponents of similar strength, I would imagine that accelerated pairings are the way to go, especially in a shorter tournament (say 7 rounds as opposed to 11).

What are the disadvantages of such a system compared to a normal Swiss? Not as suitable for deciding the outright winner?

[Stewart Reuben]

Dear Niall,
What you call the normal Swiss is probably the Seeded Swiss Pairing System. Top half v second half in rating order on all score groups throughout the event.
Its great advantage is that the system has been computerised.
Another is that it is familiar to most players.
It is efficient at determining the outright winner of a tournament where the numbers don't greatly exceed the number of rounds required for a knockout. We don't get events in Britain now where the numbers do exceed. Thus 1980 Evening Standard London Chess Congress had 250 players. 6 rounds would have resulted in 2 players scoring 6/6. As it was Bojan Kurajica won the event outright with 5.5/6 meeting the strongest opposition.

An advantage of acceleration is that there are fewer mismatches. I define those a between players rated more than 40 grading points apart or 320 rating points. You can avoid that totally in round 1, but there will be some in later rounds.

It is possible that, at the end of the tournament, a leader has to meet a much lower scoring opponent. e.g. in the recently completed 42 player, 9 round CCC Masters, Vakhidov had to play Batchelor. This often happens in the 11 rounds British Championship. Thus there is a mini-max optimum number of rounds for the number of players from that viewpoint.

In a Seeded Swiss players around the middle may suffer from the 'bouncing' effect. Thus, in Bermuda in 2001 I met a 1400 player with white in round 1 and duly won. Round 2 I met a GM with black and lost. Round 3 I met a 1600 player with white and won. Round 4 an IM with black and lost. Round 5 an 1800 player with white and won, thus scoring 3/5. That was not very satisfactory.

The British Championship did use Accelerated Pairings for a few years. The purpose was to ensure players had the opportunity to get title norms by meeting the correct mix of title opponents. But that wasn't a prime objective of the British.

Have no doubt, Acceleration is more efficient at determining the winner correctly than a Seeded Swiss. The seeded system is very poor at determining middle rankings.

Some older people prefer to use Accelerated Pairings because they enjoy the fact that they do them, not a computer. I did not use them at the CCC Masters because it would have made the event totally dependent on me or David Sedgwick. But it would have been quicker than by computer because of inexperience with Swiss Manager.