Need quick help with a tourney movement

[luis]

I'm helping organize a local tournament here

Pairs, IMP scoring.

About 40 pairs.

We have 2 sessions for this tournament, about 4 hours each one.

What format do you suggest? I mean 1 zone? 2 zones? One Mitchell and 1 howell? Any help will be appreciated.

[hrothgar]

Comment 1: Your choice of movements depends enormously on the sophistication of your players? If your players are capable of following directions and not fouling a movement, I'd recommend using some combination of Howell movements. In practice, I suspect that its easiest to simply use a Mitchell, decreasing the accuracy of the comparisons in order to remove risk and minimize the annoyance factor.

 

Comment 2: Its very difficult to recommend a movement for "about" 40 pairs, since the choice of movements is a function of the number of pairs. For example: If I were asked to recommend a movement for exactly 40 pairs and were allowed to run a Howell, I'd prefer to use

 

Two 3 table Howell movements (2 x 12 = 24 players competing over 25 boards)

One 4 table Howell (1 x 16 players = 16 players competing over 24 boards)

 

(If you prefer you could substitute a 6 table Howell

 

This is a very attractive movement. No sitouts. Every pair competes against every player in their section. Every pair plays the same number of boards against each of the other pairs. You can't really make accurate comparisons across sections meaning that its somewhat dicey selecting an overall winner, however, this scheme localizes the uncertainty.

 

In a similar fashion, 36 pairs is also quite easy to support: You can set up one 4 table Howell playing 27 boards, combined with one 5 table Howell playing 28 boards)

 

Life gets more complicated when the number of pairs isn't divisible by 4 and a LOT more complicated when it isn't divisible by 2. In this case, you need to start worrying about sitouts and all sorts of nastiness.

[MickyB]

Richard - unfortunately, it is normal to have only two pairs sit at any one table.

[hrothgar]

Richard - unfortunately, it is normal to have only two pairs sit at any one table.

Sorry, I worked things out for 40 players, not 40 pairs...

 

One dirt simple option is simply to double the number of sections

[luis]

We'll have from 40 to 44 pairs since the inscription is not closed yet, will be closed today.

We have two sessions as I described, the score has to be IMPs and the organizers want each pair to play if possible against every other pair.

The first idea was to split the pairs in 2 sections and do a Mitchell one day and 2 Howells the second day but the problem is that for 22 tables we can't do a Mitchell with only 1 board per round to play 22 boards that will not be accepted.

So far I think about a Mitchell of 14 rounds of 2 boards and you don't play against 7/8 pairs, we'll have to make sure the categories of the pairs that you don't play are more or less the same for every pair.

Any better ideas?

 

Luis

[Gerben42]

2x 13 rounds of 2 boards

 

40 Pairs:

 

13-table Mitchell, 7-table Mitchell (one pair plays Howell twice)

 

42 Pairs:

 

13-table Mitchell, 8-table Howell (three pairs play Howell twice)

 

44 Pairs:

 

14-table Mitchell (with skip so 13 rounds maximum), 8-table Howell (two pairs plays Howell twice)

 

the second day

 

I hope you mean 2nd session :)

[luis]

2x 13 rounds of 2 boards

 

40 Pairs:

 

13-table Mitchell, 7-table Mitchell (one pair plays Howell twice)

 

42 Pairs:

 

13-table Mitchell, 8-table Howell (three pairs play Howell twice)

 

44 Pairs:

 

14-table Mitchell (with skip so 13 rounds maximum), 8-table Howell (two pairs plays Howell twice)

 

the second day

 

I hope you mean 2nd session :)

Each session is one night :-) from 8pm to 12am aprox.

 

Let's see:

44 pairs, 14 table Mitchell (13 rounds) and then an 8 table howell.

So the first day/session pair #1 will play against 13 pairs in the mitchell, then what do you do in the 2nd session/day ?

Is this the most fair movement we can use?

Thanks for the help!

[hrothgar]

We'll have from 40 to 44 pairs since the inscription is not closed yet, will be closed today. We have two sessions as I described, the score has to be IMPs and the organizers want each pair to play if possible against every other pair.

TANSTAFL is a "famous" acronym from Science Fiction. This stands for There Ain't No Such Thing As a Free Lunch. The acronym is used in a variety of situations. Its often commonly used to suggest that you often can't achieve improvements in one area without sacrificing something in another.

 

Let's assume for the moment that you have 40 pairs and the tournament organizers insist that ever pair is matched for an equal number of boards against every other other pair.

 

You'd be faced with the following options:

 

1. 39 "rounds" of 1 boards each = 39 boards

2. 39 "rounds" of 1 boards each = 78 boards

3. 39 "rounds" of 3 baords each = 117 boards

 

Needless to say, the more boards that each pair plays against one another, the more accurate the comparisons. One board rounds are something of a crap shot. If the players get dealt either a completely flat or completely wild board, you won't have a valid comparison. Two board rounds are better, three board rounds better still.

 

Unfortunately, I suspect that you are stuck with a time constraint. Ideally, we'd like players to compete across and infinite number of boards. In practice, we have them for 1-2 four hour sessions. Equally significant, round changes are "expensive". You burn a lot of time as people swap seats. (Moreover, movements can be especially difficult using a Howell). In short, I doubt that time considerations will permit you to use 2 board rounds. As I noted earlier, one board rounds are completely random and result in an incredibly ineffiecient use of time due to the all the round changes.

 

Your tournament sponsors can complain about this all they want, however, you're dealing with physical constraints. This isn't the sort of thing that can be avoided through chicanery or great organizational skill.

 

Accordingly, you're going to be left with a system where the number of comparisons is incomplete. Some pairs aren't going to compete against one another. Accordingly, whatever movement you select is going to be imperfect.

 

There are a lot of different ways to deal with the problems associated with imperfect movements. One option is to muddy the waters. Run a bunch of Mitchell movements, switch pairs between sections, switch pairs from North South to East West. You won't increase the number of comparisons, but you'll confuse everyone and do a much better better job hiding the flaws in the system. As I noted earlier, I prefer to confront the issue head on.

 

Create multiple sections. Each section will use its own fully meshed Howell movement. You won't be able to perform valid comparisons between sections, however, the comparisons within sections will be as very accurate. Accordingly, you should avoid an award format based on selecting an overall winner, preferring instead to allocate masterpoints equally for the winner of each section.

 

If the tournament organizers complain, you need to have a serious discussion with them. They need to specifically decide if they want run an "accurate" tournament or simply pretend that they are running an accurate event. If they're actually concerned with identifying the best pair in the event and are unwilling to spend enough time to run a perfect movement, the best option is a barometer. If they're happy to pretend, just run Mitchells

[luis]

Yep seems it is impossible to do it perfectly so I wonder which is the best compromise solution.

They need only 1 pair as winner because it's a sort-of national competition (sigh)

Why do you sugest a barometer? What is good with a barometer?

I keep listening to ideas and suggestions so far the best I have is:

 

A 22 tables Mitchell with 14 rounds of 2 boards each (28 boards) the first day.

You don't play against 7 or 8 boards. So we have to make sure the seeding is very accurate to make sure the pairs that each one don't face are of the same weight.

Then 2 11-tables howells playing against 21-22 pairs (2 boards each round).

 

Luis

[Cascade]

For 4n pairs you can always create a movement over two sessions where you play a 2n-table Mitchell in one session and two n-table Howells in the other session. The NS pairs from the Mitchell form one Howell and the EW pairs form the other Howell. This way everyone will play every other pair.

 

For 40 pairs you would need a little longer than your allocated time. You would play a 20 table Mitchell for 40 boards 4hr 40 min at 14 minutes per round and two 10 table Mitchells for 38 boards 4 hr 26 min at 14 minutes per round.

 

You can improve the movement for a one winner field by arrow switching the Mitchell.

 

If you have 2n pairs (which is not divisible by 4 e.g. 38 pairs) then there are at least two solutions for a similar movement.

 

n-table Mitchell - 19 tables for 38 pairs and two 10-table Howells and either there is a sitout in Each of the Howells OR you can dove-tail the Howells so that the two sitout pairs play each other. This can be arranged so that pairs that start together in the Mitchell will meet again in the Howell. By skipping the first round of the Mitchell then you can run the movement with no sitout and only meeting each pair once.

[Gerben42]

So the first day/session pair #1 will play against 13 pairs in the mitchell, then what do you do in the 2nd session/day ?

 

You have 3 groups of 14 pairs:

 

1 - 14

101 - 114

201 - 214

 

and two other pairs: 301 - 302

 

1st session:

 

1 - 14 vs 101 - 114 in Mitchell

 

and 201 - 214, 301, 302 in Howell

 

2nd session:

 

1 - 14 vs 201 - 214 in Mitchell

 

101 - 114, 301, 302 in Howell (such that 301 & 302 don't play eachother)

 

Arrowswitch a couple of rounds before the end in the Mitchell to make it fair and get one winner.

 

Ideally you'd want to play 1.5 boards against each pair given your time constraints, no such thing I'm afraid...

 

If you play with 15 board sets you will have a virtual 15th table so you can play the same movement with 28 total rounds.

[luis]

So 1-14 never play agains 1-14, why is this better than a big Mitchell and then 2 Howells? I'm not arguing just asking you because it seems you know this quite well :-)

Thanks for the answers Gerben.

 

Still looking for the best solution :-)

[Cascade]

Playing a Mitchell and a Howell in each session is better if you do not have enough time to play against every other pair. This way you play against 2/3 of the field (roughly). If the field is seeded into three even groups then this should work out ok.

[Gerben42]

Two reasons:

 

1.This way you can control who is playing against who.

 

Group 0 is not playing against itself.

Group 1 is not playing aginst group 2.

Group 2 is not playing against group 1.

 

So if you have 3 roughly even groups your tournament will be balanced. If you play Mitchell on day 1 and Howell on day 2 that's not so easy.

 

2. There are also a practical problem with a 22-table Mitchell movement and then a big Howell, and it's to do with the boards.

 

If you play in one big group you need as many board groups (I think that's the right word) as tables, in this case 22, this means board numbers 1 - 44. If you play only 28 of those, that's not very good.

 

Perhaps more experienced TDs may have a solution to that one also, but it complicates things.

[jikl]

A dealing machine and 22 sets of boards? *runs away and hides*

 

Sean

[Blofeld]

If aiming for a winning pair, can you do something like a swiss movement?

 

Even if a real swiss movement is impractical, could you run the second session so that the best pairs from each section in the first session play together in a section? Then even if you didn't have an entirely accurate total ranking it would be more accurate at the top than otherwise.

[MickyB]

It isn't necessary to try to have everyone play against similar quality of opponents.

 

Say you have a 10 table Mitchell with no arrow-switch. The NS pairs don't really compete against the EW pairs at all, but against the other NS pairs. If you add in one round of arrow-switch, the NS pairs are still competing against each other more than they are the EW pairs - the competition ratio is 7 against each pair sitting their direction, 0 against the pair that they arrow switch against, and 4 against all the other pairs sitting in the other direction.

 

I've only just got interested in movements, so I'm not yet much good at putting this into practice. If I come up with anything worthwhile, I'll post again. Have a look at John Manning's work on Chris Ryall's site if you are interested in the mathematics of movements.

[MickyB]

I think it should work to just have two eleven table Mitchells. For the 2nd session, just swap the EW lines over. Only downside is that it would probably require some fiddley arrow-switching - I'll think about it more tomorrow.

[Gerben42]

Not sets of boards, board groups.

 

Board group 1: Board 1 - 2

Board group 2: Board 3 - 4

etc.

 

Luis will need two copies of each board, so two board sets.

 

Micky: Two 11-table Mitchells won't work since you can only have 22 rounds compared to my 28 rounds.

[MickyB]

Micky: Two 11-table Mitchells won't work since you can only have 22 rounds compared to my 28 rounds.

Why does that mean that my suggestion wouldn't work?

 

I'm not saying it does, just that I can't see where you are coming from :D If the movement is balanced, everyone plays every board, and 33 boards (3 board rounds) is about right for a 4 hour set, then the movement looks pretty good to me.

[Gerben42]

The side condition was that they wanted to have as many opponents as possible, which means 2-board rounds.

[MickyB]

Ah - I assumed that was because of the perception that if you didn't play against a pair, you weren't competing against them.

 

IMO 3 board rounds is much better if the movement is well balanced, because it means more bridge and less moving.

[Limey_p]

The "web movement" is way to move boards so that a single large mitchell section will be played, and everyone plays the same boards. The *only* link to it I have seen is http://www.bridgeguys.com/WGlossary/GlossW.html, scroll down to "web movment". I understand you need two or three sets of boards for 20-25 tables.

 

The ACBL's Delaware unit (www.unit190.org) frequently use this movment for their monday evening game - it is fun and seems fairer than playing within smaller sections.

 

So my suggestion for a two session tourny would be:

 

First session: One section, web movment, 2 bds/round.

 

Second session: two (partial) howells, one the original NS field & one the original EW.

 

Another option could be to run a "swiss pairs" - I've never seen one in the USA but I believe they are common enough in other parts of the world. Anyone care to describe this option for Luis?

[mike777]

I am a bit confused what does the Math have to do with choosing the "best movement"?

 

As usual it seems semantics are most important..what does best mean?

 

Perhaps "Marketing" to your paying customers is best? "know your customer"

[awm]

Swiss is a pretty random way to run things. The problem is that if you're continually playing against comparable opponents, you expect to have average rounds. If there's one pair (or team) that's substantially the best, they should still come out on top, but swiss events don't do a good job of accurately ranking the lower places.

 

Of course, one reason swiss teams are very popular is that they are random! Apparently there used to be more Board-a-Match events at regional tournaments here in the US, and as BAM is not very random, the same teams kept winning. People didn't like this, and so they switched to the much more random Swiss Teams format...

 

Anyways, if you're looking for a result that's somehow fair and accurate I recommend against Swiss Pairs. The only example of Swiss Pairs in recent times in the United States was the junior team trials a year and a half ago, and everyone involved with that was convinced it was pretty random.