Some questions about SwissSys pairings and colour distribution

[Jordan S. Berson]

Here are a couple of scenarios that happened in the 2009 Newfoundland Open regarding SwissSys pairings. I'd like to know what players and TD's think:

Scenario 1: There are 10 players. Two had byes in the first round; one was requested, so another player had a full-point bye. There were no upsets and no draws in the first round. After one round, there are five players with 1.0, one with 0.5, and four with 0.0. The two top-rated players are paired in round two, even though there are five players with 1.0/1.

Question 1a: Is it more important to ensure that all players get even colour distribution, or is it more important to pair 1 vs. 3, 2 vs. 4?

Question 1b: If colour distribution is more important than point ranking, then couldn't round two be paired as 1 vs. 4, 2 vs. 3 to avoid a confrontation between the top two players so early in the tournament?

Scenario 2: In round three, the top player is paired against the player who had the first-round forced bye (both players have 2.0/2). Both players had played Black in round two. SwissSys paired the top player with Black again, even though his opponent had a bye in round one.

Question 2: Shouldn't the top player automatically switch colours as long as his opponent is not forced to switch?

The fact that this is a five-round tournament means everyone will have more of one colour than the other, except for the two players that had byes. With only 10 players, it is impossible to avoid having two of the same colour in a row for at least 8 or 9 of the players, so is it really so important to avoid repeating colour in round two or three, and ignoring other pairing rules?

I remember in FQE tournaments, round two is always paired top vs. top of the bottom half, even if both players just had the same colour. I had Black in my first two rounds at this year's Quebec Open, and again in round four, but that's how the pairings should work. However, in CFC tournaments, it seems pairing for colour far outweighs all of the other basic pairing rules.

Any insight and/or rules would be greatly appreciated. With the CFC website still out of commission, I cannot check the rule book for Swiss pairings.

As a footnote, I would like to congratulate Alick Tsui for organizing another fine NL Open tournament. The above post is not a complaint about the pairings, I'm just curious as to what the actual rules are. I highly doubt that any different colour pairing would have boosted my score beyond 4.0/5, especially since my loss was from a rather strong position with White. The only thing that could have helped me is better nerves... and maybe a more cooperative opponent in round 4 ;)

[Stephen Wright]

Re: Some questions about SwissSys pairings and colour distribution

Question 1 - hard to answer without knowing exactly who had the byes (and therefore are not concerned about colour), but in general pairings should be done top half to bottom half unless there is a good reason not to. So your 1b pairings are perfectly acceptable, assuming the colours work.

Question 2 - not necessarily, the CFC rules state that equalization takes precedence over alternation.

Not sure about your comparison of FQE vs. CFC rules; it is my experience that SwissSys places a much higher emphasis on colour issues than the CFC handbook does, and thus produces pairings which not infrequently break CFC rules.

[Jordan S. Berson]

Re: Some questions about SwissSys pairings and colour distribution

Question 1 - hard to answer without knowing exactly who had the byes (and therefore are not concerned about colour), but in general pairings should be done top half to bottom half unless there is a good reason not to. So your 1b pairings are perfectly acceptable, assuming the colours work.

Question 2 - not necessarily, the CFC rules state that equalization takes precedence over alternation.

Not sure about your comparison of FQE vs. CFC rules; it is my experience that SwissSys places a much higher emphasis on colour issues than the CFC handbook does, and thus produces pairings which not infrequently break CFC rules.

Steve,

Thanks for the insight. I have a copy of SwissSys, but I have not really had a look at the options. My understanding is that the weight of different criteria can be adjusted to suit local rules. Do you know anything about that?

Thanks again, Jordan

[Stephen Wright]

Re: Some questions about SwissSys pairings and colour distribution

I have tried tweaking the SwissSys settings, but have yet to find any that avoid what I regard as anomalous pairings. If such settings do exist the CFC should advertise and promote their use.

As an aside I suspect many/most TDs these days simply press the button and uncritically accept what the machine produces. In one sense there's nothing wrong with this, as long as the program is internally consistent, but the pairings don't match what's laid down in the CFC handbook. The other problem is we are getting a generation of new TDs who rely so much on SwissSys that they never learn how to pair a section manually and thus wouldn't know if the pairings were 'correct' or not.

[Christopher Mallon]

Re: Some questions about SwissSys pairings and colour distribution

We did have three or four instances in the Closed where Swisssys would flip the colours on two people, in order to maintain perfect colour alternation for one of them. The problem is that this is not correct of course and we had to manually correct the pairings.

Most TDs probably would, as you say, just click Pair and go with it - and most players probably wouldn't even notice or know the difference. Hopefully it's something that can be fixed in a future version of SwissSys.

[Jordan S. Berson]

Re: Some questions about SwissSys pairings and colour distribution

Thank you Steven and Christopher, you have both confirmed my suspicions.

While I did not necessarily agree with the pairings, I asked the TD about them, but I did not dispute them. There would be no point in getting upset over it, especially where I was not of the belief that a change in colour would change my winning chances in any particular round, nor would playing the #2 player in round three as opposed to round two. In fact, I beat the #2 player with Black, so you can't ask for much better than that ;)

I do agree there should be some form of agreeance between all CFC TD's that the pairings will be in accordance to CFC law, and not just what SwissSys calculates based on different weights. However, getting such an agreeance might not be so easy...

Jordan

[Steve Douglas]

Re: Some questions about SwissSys pairings and colour distribution

Hi Jordan:

There were two things that I noticed about SwissSys. One is fairly obvious but the other is kind-of subtle. The first is that USCF rules allow for 3 Blacks in a row, which SwissSys will happily do if you allow it. The default is usually to not allow it.

The second issue is that in most tournaments where SwissSys is used, TD's don't bother to specify the number of rounds. This can lead to odd pairing/colour issues in the second-to-last and last rounds, since SwissSys will assume that whatever irregularities it introduces will be corrected with a later pairing in a future round that never happens. From what I can tell SwissSys uses a sort of "look-ahead" with predicted results to try to preserve clean pairings for the last round. If it doesn't know when that is ... well then tomorrow is always tomorrow and never today.

Steve

[Jordan S. Berson]

Re: Some questions about SwissSys pairings and colour distribution

Hi Jordan:

There were two things that I noticed about SwissSys. One is fairly obvious but the other is kind-of subtle. The first is that USCF rules allow for 3 Blacks in a row, which SwissSys will happily do if you allow it. The default is usually to not allow it.

Steve,

Are you sure about that? That seems like a huge error...

The second issue is that in most tournaments where SwissSys is used, TD's don't bother to specify the number of rounds. This can lead to odd pairing/colour issues in the second-to-last and last rounds, since SwissSys will assume that whatever irregularities it introduces will be corrected with a later pairing in a future round that never happens. From what I can tell SwissSys uses a sort of "look-ahead" with predicted results to try to preserve clean pairings for the last round. If it doesn't know when that is ... well then tomorrow is always tomorrow and never today.

Steve

Noted... thanks for the insight!!

Jordan

[Steve Douglas]

Re: Some questions about SwissSys pairings and colour distribution

Are you sure about that? That seems like a huge error.

Hi Jordan:

I just checked an older version of SwissSys (version 5) and the "USCF defaults" allow three consecutive colours (it's a check-box in the pairing rules). The "FIDE defaults" do not naturally. There are some other USCF oddities regarding colours such as *not* doing colour equalization on even rounds under certain circumstances. Thus it's possible to play a six-round event and get two whites and four blacks (seen it happen).

Steve

P.S. My information is obviously a couple of years out of date. The USCF may have changed their rules, but I haven't heard about it.