Six Board Swiss Matches

[TimG]

I received an e-mail today that suggested some regionally rated Swiss Team events at NABCs now consist of eight six-board matches.

 

Can someone provide numbers for the difference in reliability of a six-board result as opposed to a seven-board result? How does this affect the reliability of the overall event results?

 

I suspect that for the event, this change would be similar to the difference between a 56-board and a 48-board KO match, but it must be different because of the VP factor.

 

Thanks.

[ArtK78]

Given that many regional swiss teams now consist of 7 7-board matches, a more relevant comparison might be to compare the results of 7 7-board matches scored at VPs as opposed to 8 6-board matches scored at VPs. How you would do so is beyond me, but I am sure someone will work it out.

[barmar]

The play-through swiss on the last day in Reno was 7x7, but I don't recall playing in any 8x6 events. There may have been some at previous NABCs, though. Those might have been the same tournaments where regional pair games were only 24 boards, rather than the traditional 26. My understanding is that the host district dictates how regional events are played (they also get to specify whether VP events use the 20 or 30 scale).

[TimG]

Yes, I believe you are right that the host District generally specifies the formats used in regionally rated events at NABCs. The chatter I heard was from the last Fall NABC, not the Spring NABC held recently in Reno.

 

Still, I'm curious about the differences in expectation between the different formats.

[jeremy69]

The EBU have held tournaments at 6 7 8 & 9 boards. Most Swiss Team Events now are 7*7 boards. 8 and 9 took slightly longer than was wanted and playing 7*7 gave not only 1 whole extra board but one extra opportunity to earn master points which is important to a lot of swissers. In theory 6*8 gave even more opportunity but it was slower because of extra changeovers, more difficult in smaller fields as you searched harder for opponents with a similar score and slightly more random because the match was shorter. I'm not sure any of the decisions were arrived at scientifically but everyone seems fairly happy.

[hrothgar]

Given that many regional swiss teams now consist of 7 7-board matches, a more relevant comparison might be to compare the results of 7 7-board matches scored at VPs as opposed to 8 6-board matches scored at VPs.  How you would do so is beyond me, but I am sure someone will work it out.

1. Look at the results of a few tournaments. Try to make some reasonable assumption about how the strength of the field is distributed. (Normal distribution, bimodal, what have you)

 

2. Create a set of virtual teams. The strength of each team is modelled as a draw from the previous distribution.

 

3. Have the teams "compete" against one another in a virtual tournament using format 1. (7 rounds of 7 boards each).

 

4. Have the teams "compete" against one another in a virtual tournament using format 2. (8 rounds of 6 boards each)

 

5. Compare the results of the virtual tournaments with the "objective" truth (The known team strengths that were observed in advance of the tournament).

 

6. Repeat a few thousand times and see whether there is a statistically significant difference between the two methods. If there is, see which method works best.

 

The trickiest issue is (probably) the best way to evaluate accuracy of the two systems. You need to make a precise decision what you want to measure:

 

Do you care how accurately you are ranking ALL of the teams?

Alternatively, are you only interested in how accurately you rank the top "N" teams?

 

I would argue that the master point allocation schedule might be a reasonable way to determine which set of team you care about. For example, if the top 5 teams receive Master Points, then you should be most interested in how accurately the tournament can rank the top 5 teams.

 

One issue that does worry me somewhat is the relationship between round length and the size of the field. If you impose no-playback restrictions, you might run into some weird issues if the round size is too large relative to the field size.

 

As an aside, I've always thought that master point schedules should be linked to the "accuracy" of the tournament format. Lets assume that you have a format that is

 

1. Very accurate in identifying the best team

2. So-so in identifying the second best team

3. Miserable at ranking teams 3-8

4. Hopeless otherwise

 

Then:

 

First place should pay a lot

Second place should pay a decent amount

The awards for 3-8 should all be about equal.

[TimG]

I suspect the masterpoints awarded to each team over a sample of tournaments would be a reasonable way to measure the results.

[Siegmund]

It'll be quite close between 8x6, 7x7, and 6x8. The last being least overswissed has something to recommend it, as well as playing substantially faster with fewer changes.

 

As far as the comments I'd make to a tournament sponsor -- for all three of the above my answer is the same: what a ripoff, only 48/49 boards, instead of the 52-56 I'd otherwise get. Your only two options I find acceptable are 6x9 and 8x7.

[mycroft]

I can provide evidence that the one-day Swiss we played the day before the Vanderbilt was 8x6 (two sets of 4); all one-session Swisses have been 4x6 for a while.

 

The last Sunday, yes, was 7x7 playthrough.

 

And yes, I've never been fond of the "play less" solution to slow play, either. Why do you ask?

[Mbodell]

I was a big fan of the fact that the Reno regional pairs were generally 26 boards instead of the 24 we've seen at some previous nationals.

 

I also like the 4x6 swiss matches and 8x6 swiss matches with 20-VP scoring rather than the 30-VP scoring some previous nationals had used. The play through final day swiss being 7x7 felt better than the 6x8 from San Diego.

 

So overall I though the Reno event was well organized.

[shintaro]

 

And yes, I've never been fond of the "play less" solution to slow play, either.  Why do you ask?

;)

 

Just intrigued as to why anyone could even think that playing less boards per match could ever cure slow play :)

[keylime]

I am much more a fan of the 7x7 format than the 6x8 format, but I would personally angle towards the 8x7 format.

[ggwhiz]

This is 5-6 years old but maybe what the whole deal is about?

 

Memo to the ACBL

 

Many of the reasons for the decline of our sectional tournaments have been discussed but an important one has been totally overlooked.

 

It relates to the manner in which the Sunday Swiss events are run. They are starting earlier than ever before and are run as sprints to the finish line with barely enough time between sessions to catch your breath. In many Units, the event has been reduced to seven matches with no breathers at all.

 

There is a rumor floating around that the Directors have hired a professional lobbyist to have the number of boards played in each of these matches reduced to two.

 

While the current format had its place and contributed to the boom of the nineties, the economy is back in the tank, a place and time that I believe your membership wishes to return to. The early finish has raised some very serious stamina issues and there is less reason than ever to be functional on a Monday morning.

 

Please don’t insult me with the argument of our ageing membership demographic. Recent medical advances clearly show that blown gaskets can be repaired as never before, with speed and precision.

 

As to our shrinking demographic, this format has chased an entire segment of bridge players away. They are the Saturday night players who typically suffer from flu-like symptoms before noon on any given Sunday.

 

Memo from the ACBL

 

We have noted your position regarding Sectional tournaments and the Sunday Swiss in particular. I would like to thank you for your contribution to an on-going debate.

 

You may be right about the timing of these events and the effects of their finishing times on attendance. We are not able to finance a return to the old ways as it would require a significant raise in pay for the Directors to do that much of nothing once again.

 

It is true that they are well and professionally represented and we are currently studying a compromise proposal that addresses most of your concerns.

 

Yes, the Saturday evening games are also in trouble but we can achieve a more appropriate finishing time by starting the Sunday event immediately after the Saturday afternoon game and playing it as a speedball play through.

 

This new event would be run as two board knockout matches and result in the kind of staggering finishes that used to be so popular in our midnight games. Based on current attendance numbers, the whole Sunday event can be wrapped up by 8:00 pm on Saturday and when the anticipated surge in attendance is realized, we will bracket it so as to produce multiple winners and a conclusion by 7:00 pm.

 

The efficiency gains for our Directors and the realization of your stated goals a full day early is a classic win-win situation and I thank you once again for your suggestions.

[Bbradley62]

I was a big fan of the fact that the Reno regional pairs were generally 26 boards instead of the 24 we've seen at some previous nationals.

When did 24-board sessions become common? I've been away from tournaments for 10 years and never heard of a regional event with fewer than 26 boards. And Swiss Teams, even on getaway day, was always 8 matches of 7 boards.

 

IMO Swiss Team events (even those using VPs) should be designed to have the number of rounds it takes to come down to one undefeated team: 4 rounds for 9-16 teams, 5 rounds for 17-32 teams, 6 rounds for 33-64 teams, 7 rounds for 65-128 teams, etc. The number of boards per round should be whatever it takes to make the desired total number of boards per day. (I'd suggest 54-56 boards for anywhere from 4-9 rounds, but that's just me...)

[ArtK78]

Back in the days of win-loss non-flighted Swiss Teams, there were some very large regional swiss team events at a few of the North American Championships (then referred to as Nationals), most notably Las Vegas, in which there were ties for first between three undefeated teams after 8 matches! I am sure that is one of the reasons why Victory Points were introduced in the regional swisses at Nationals shortly thereafter. Flighting came into vogue at about the same time which would have also solved the problem by reducing the size of the event.

 

On the other hand, I once had a clear win with 7 3/4 wins out of 8 (a win by 1 or 2 IMPs was a 3/4 win) in a 400 team open swiss teams at the Chicago Nationals in 1980 (I think it was 1980) when the other team with 6 3/4 wins entering the last round beat the only undefeated team by 2 IMPs in the last round after we had beaten our opponents. So the top three teams had 7 3/4 wins, 7 1/2 wins and 7 1/4 wins out of 8 matches.

[ggwhiz]

Before my time a bit but If I remember correctly, the Toronto Regional used to run a 9th round as needed for the few undefeated teams to resolve any ties.

[barmar]

I was a big fan of the fact that the Reno regional pairs were generally 26 boards instead of the 24 we've seen at some previous nationals.

When did 24-board sessions become common? I've been away from tournaments for 10 years and never heard of a regional event with fewer than 26 boards. And Swiss Teams, even on getaway day, was always 8 matches of 7 boards.

I don't know if it's "common" yet, but the two Nationals I attended in 2009 (summer and fall) both played only 24 boards in regional events. But the spring National this month was back to 26. I think the hosting district decides this, so it can vary from tournament to tournament.

 

And I think the play-through Swiss on the last day has been 7x7 for quite a few years.

[Mbodell]

I was a big fan of the fact that the Reno regional pairs were generally 26 boards instead of the 24 we've seen at some previous nationals.

When did 24-board sessions become common? I've been away from tournaments for 10 years and never heard of a regional event with fewer than 26 boards. And Swiss Teams, even on getaway day, was always 8 matches of 7 boards.

I don't know if it's "common" yet, but the two Nationals I attended in 2009 (summer and fall) both played only 24 boards in regional events. But the spring National this month was back to 26. I think the hosting district decides this, so it can vary from tournament to tournament.

I'm pretty sure that it is at the tournament organizer's choice. And since the side events at nationals are basically a concurrent regional I think it is up to the district to decide. We had some folks here in district 21 tossing around doing hand record for the regional swiss in the future now that we have several dealing machines in the area, and I think decisions like that are mostly up to the district in charge.

 

And I think the play-through Swiss on the last day has been 7x7 for quite a few years.

 

It was 6 by 8 with a 30 VP scale in San Diego last fall, at least the A/X on the last day. I don't remember for sure which DC was, but it may well have been 7x7 with a 20 scale.

 

IMO Swiss Team events (even those using VPs) should be designed to have the number of rounds it takes to come down to one undefeated team

 

I don't agree with that. I think there should be more rounds than that, especially with VP. Most events aren't just picking a winner, and extra rounds help make sure that the winner, and top few places, are more likely to be reasonable. I think something more like floor(1.5 * sqrt(n)) rather than ceiling(sqrt(n)) would be better.

 

So with 4 teams you get a round robin instead of only 2 rounds. With 16 teams you get 6 rounds instead of just 4. With 64 teams you have 9 matches instead of just 6.

 

I mean there is no reason why a swiss needs to be so like a KO. And especially with VP you aren't guaranteed that there will be only one undefeated team nor that the undefeated teams will have more VP than a team with a loss. So you may as well embrace that it can be more like a RR and play more rounds.