Calling all format wizards

[Jlall]

Just curious.

 

The Texas GNT is a 2 day affair. Assume each day has to have 52 boards, and it has to be played on a Sat/Sun (I realize that playing matches over several weekends etc would be preferable, not really looking to hear anything on that).

 

The goal of the event is for 1 team to win and advance to the nationals. Assume the goal is for the best team to win as often as possible.

 

If there are 4 teams then obv you just play 2 52 board matches, easy game. However there is a distinct possibility of a 5 team event. If that happens the conditions of contest stipulate a 52 board round robin to cut to 4, then a 26 board semi and a 26 board final. I assume there will be carryover, and that the winner can select his opp from 3/4.

 

That format seems retarded to me. I'm curious what the best format is though. Playing a round robin and cutting to 2 for a 52 board final seems like a steep cut. Playing a full round robin has it's problems, namely beating up on the bad teams becomes very important and ideally the best team is the one that can beat the other good teams.

 

Maybe something like day 1 is a round robin to drop 2 teams, and day 2 is a round robin between the top 3 teams? Maybe something else completely since I'm not creative/know nothing about formats etc?

 

Thanks for your help!

[Gerben42]

Hi Justin,

 

That's a nice problem you have there. I think the fairest and yet most exciting format would be something like:

 

Day 1: Round robin: four matches on a VP scale, reduce to 3 teams.

 

Carry-over 2/3 of the VP to day 2.

 

Day 2: Double round robin (four 13-board matches) with the 3 remaining teams.

 

Example, after day 1 we have (using international scale the average is about 60)

 

A. 78

B. 72

C. 60

D. 50

E. 30

 

First three advance, times 2/3:

 

A. 52

B. 48

C. 40

 

Now play twice against each other team on the same VP scale.

[jjbrr]

Online.

 

Matchpoints.

[Jlall]

Hi Gerben,

 

I was thinking maybe dropping your score completely against the 2 teams that are cut (doesn't really matter how well you do against those teams imo), and having full carryover against teams that are not cut.

 

This does create some possibility for dumping scenarios where you dump to a team which hurts you for the next day but might be beneficial if it knocks out a team that is crushing you, but it's unlikely that a team would be crushing you and in the bottom 2, while you're so comfortably in the top 3 that you can dump so I don't think it would be that bad.

 

How do you think this compares to your idea?

 

Also, you recommend VPs for the 3 way. This implies to me that you think if one team wins 2 matches by 1 imp, and another team wins 1 match by 100 and loses the other by 1 (extreme example), that the latter team should win. Is there any merit in your mind to the team that wins both matches being the better team, even if they have a lower + imp quotient?

 

In my head I had been thinking full imp carryover from day 1 to day 2 which the other scores dropped, then if one team wins both matches they are the winner, if all are 1-1 it goes to imp quotient.

 

Do you think this would make it less likely that the better team wins?

[Echognome]

I know others have modeled these things, but it seems to me better to use Day 1 to get down to 2 teams and Day 2 to have a 52 board final. The tradeoff being that you have a little more variability in Day 1 that the best team could get knocked out, but a lot of equity is gained on Day 2 that the best team will prevail.

[Phil]

GNT Conditions of Contest require 64 boards I believe.

 

5 is a tough number. I would say play a 5 x 5 swiss to eliminate one team, but carryover the margin to the next round and maintain seed positions. Play 32 (or 26) in the semis and 52/64 in the final.

 

D22 has some guidelines for these events here.

[Apollo81]

Let T be a team competing in the current-year district level quals that is composed entirely of team members from the winning squad in the previous year. If T exists, it automatically gets the top seed. Seed remaining teams using whatever method you prefer. Let T1 be the top seed.

 

5 teams: T2-T5 and T3-T4 play head-to-head 26 board matches. The winners play a 52-board 3-way match with T1 eliminating one team. The remaining teams play an additional 26 boards to determine the winner of the event, with full carryover from the 3-way match.

 

6 teams: Similar to 5, but T1 plays T6 in the first round rather than getting a bye.

 

7 teams: Three head-to-head 26 board matches. Winners earn the seed of their defeated opponent. T1 plays their choice of the two lowest remaining seeds in the second round, which is two head-to-head 26-board matches. 52-board final.

 

8 teams: Similar to 7, but T1 plays T8 in the first round.

[Jlall]

GNT Conditions of Contest require 64 boards I believe.

This is dictated by the district.

 

Anyways obv 64>52, as I said assume 52 and 2 days are forced.

[Jlall]

5 teams: T2-T5 and T3-T4 play head-to-head 26 board matches. The winners play a 52-board 3-way match with T1 eliminating one team. The remaining teams play an additional 26 boards to determine the winner of the event, with full carryover from the 3-way match.

Meh I think T2-T5 would object to this so it would never fly but it's an interesting idea.

[Jlall]

Also I just realized another interesting thing about the way our district currently has it set up is that it would be very beneficial for everyone but the best team to "encourage" some random 5th team to join the event if only 4 are going to play. I guess that's the case in any format though, kinda sucks.

[kenrexford]

I wonder why no one ever runs an "IMP Pairs" type of approach in the early round(s). You run through the IMP Pairs, and then add up the IMP's of the N-S pair and the E-W pair for the team score.

 

That has a downside of not being able to bid/defend as if running against the specific opponent team you are trying to defeat, but it has a purity about not getting hammered as much by a freak result by some fluke team.

[Apollo81]

Here is another creative suggestion for 5.

 

R1: T2-T5 and T3-T4 play 26-board head-to-heads.

R2: Winners play a 3-way with T1. Losers play a 26-board head-to-head.

R3: Winner of 3-way chooses from loser of 3-way and winner of R2 head-to-head. Two 26-board head-to-head matches.

R4: Two remaining teams play 26-board head-to-head.

 

IMPs from previous matches are carried over where applicable. (maybe half IMPs from a 26-board previous match and full IMPs from a 13-boarder)

 

 

This has the advantage of collecting more entry fees, much like a 5-team RR cutting to 4.

[cherdanno]

A round robin of 5 followed by a round robin of 3 (with some kind of carry-over) sounds reasonable. However, I am not sure whether it's actually better than a round robin of 5 followed by a knockout match.

How about a day of round robin of 5, followed by half a day of 3-way, followed by 26 boards final (all with carry-over of course)?

[Jlall]

A round robin of 5 followed by a round robin of 3 (with some kind of carry-over) sounds reasonable. However, I am not sure whether it's actually better than a round robin of 5 followed by a knockout match.

How about a day of round robin of 5, followed by half a day of 3-way, followed by 26 boards final (all with carry-over of course)?

Yeah I was thinking about your last suggestion.

 

That would be 52 boards cut to 3 teams.

26 boards of a 3 way

26 boards heads up

 

That means that first and 2nd will have played 52 boards against each other.

 

That is pretty simple and eliminates the results vs the bad teams from the 3 way/finals (I like), and allows for adequate amount of play vs the semifinalists/finalists without cutting so many teams that the best team could get eliminated from the round robin.

 

To me that seems very superior to cutting to 2 and playing a 52 board heads up with a 13 board carryover. Cutting from 5 to 2 puts too much emphasis on beating up on the weaker team and increases randomness a lot imo.

[mike777]

Just curious.

 

The Texas GNT is a 2 day affair. Assume each day has to have 52 boards, and it has to be played on a Sat/Sun (I realize that playing matches over several weekends etc would be preferable, not really looking to hear anything on that).

 

The goal of the event is for 1 team to win and advance to the nationals. Assume the goal is for the best team to win as often as possible.

 

If there are 4 teams then obv you just play 2 52 board matches, easy game. However there is a distinct possibility of a 5 team event. If that happens the conditions of contest stipulate a 52 board round robin to cut to 4, then a 26 board semi and a 26 board final. I assume there will be carryover, and that the winner can select his opp from 3/4.

 

That format seems retarded to me. I'm curious what the best format is though. Playing a round robin and cutting to 2 for a 52 board final seems like a steep cut. Playing a full round robin has it's problems, namely beating up on the bad teams becomes very important and ideally the best team is the one that can beat the other good teams.

 

Maybe something like day 1 is a round robin to drop 2 teams, and day 2 is a round robin between the top 3 teams? Maybe something else completely since I'm not creative/know nothing about formats etc?

 

Thanks for your help!

fwiw...the stated format does not seem that bad but if many players think so....it is worth a discussion to change it.

 

 

I don't know if the assumed goal, is the true goal given this format.

 

For instance the goal maybe to insure all teams play as many bds as possible with the least randomness.

 

Or the true goal may be to encourage as much participation as reasonable, even if that means a bit of a less chance the best team will win as often as possible.

[Jlall]

Hey Mike, I mean ours is held during a sectional and there is a flight A swiss the next day for those who don't qualify. I'm all for randomization of smaller events and increasing participation, really, but for events that qualify you for a legit national event or a world championship I think the best team winning should be the priority.

 

Either way if there is 3 teams instead of 4 teams in the second day it's just 1 team who has to play a flight A swiss instead of the rest of the event, I think maintaining the integrity of the event is a priority.

 

Obv you might be right that my stated goal is not the actual goal, and I'm obv biased in that respect.

[Mbodell]

It could be worse, you could be trying to cut from between 17 and 20 teams down to just 1 in 4 28 board sessions. Admittedly this is just for the B flight of GNT, but the last I heard the proposal is that the last session will be a 3-way over 28 boards to name one winner, which should be pretty random given that there are a number of quite close to equally good teams.

[JoAnneM]

Justin, I am posting the link to D20's Conditions of Contest. I know they have been negotiated extensively the past several years with our open flight players and they seem pretty happy with the setup. Of course we don't have the depth of talent that D16 has. But looking at different CoC's might give you some perspective.

 

Jo Anne

 

http://acbld20.org/lbc/NA_Events/D20_GNT_CoC.pdf

[gnasher]

Are five teams all that entered originally, or are they just what's left after some earlier rounds?

[TimG]

I looked up the CoC for District 25 (New England) and found that with five teams there is an all day round robin to reduce to four teams followed by all-day semi-finals and final (with the final being played at a later date). Doesn't help given your requirement that the event be completed in two days.

[Jlall]

Justin, I am posting the link to D20's Conditions of Contest. I know they have been negotiated extensively the past several years with our open flight players and they seem pretty happy with the setup. Of course we don't have the depth of talent that D16 has. But looking at different CoC's might give you some perspective.

 

Jo Anne

 

http://acbld20.org/lbc/NA_Events/D20_GNT_CoC.pdf

Thanks.

 

Gnasher 5 teams entered originally

[Cascade]

How about

 

1. Round Robin 4/5 to progress 4x7 board matches

 

2. Round Robin 2/4 to progress 3x8 board matches with carry forward from the non eliminated team

 

3. 52 board final on day two

[gnasher]

I think 3-ways are awful, especially later in the event. If the sole objective is to prduce a winner, I would do:

- A one-session round robin to reduce to four teams.

- Two one-session head-to head matches, with the round-robin winners choosing their opponents from amongst the other three. In each match the IMPs from the round-robin match between the same two teams should be carried over.

- A two-session head-to-head final, again with carryover from the corresponding match of the round robin.

 

Of course, the team that gets eliminated after one session might not be very pleased. But they should have played better.

[xcurt]

I think 3-ways are awful, especially later in the event. If the sole objective is to prduce a winner, I would do:

- A one-session round robin to reduce to four teams.

- Two one-session head-to head matches, with the round-robin winners choosing their opponents from amongst the other three. In each match the IMPs from the round-robin match between the same two teams should be carried over.

- A two-session head-to-head final, again with carryover from the corresponding match of the round robin.

 

Of course, the team that gets eliminated after one session might not be very pleased. But they should have played better.

Having the RR winner get to pick, combined with carry-over, allows targeting. Let's assume the putative best team (probably Hamman and friends in this context) qualifies but doesn't win the RR. Assume the other three qualifying teams are roughly equal in ability (might or might not be the case in practice). If I'm the RR winner, and my carry-over against the other two teams is fairly close, I might even pick one of them so that the other get's to play Hamman's team with a big carry-over.

 

Intuitively, I agree strongly about not having 3-ways.

 

Also intuitively, I agree if you play a 5->4 RR it's quite likely the eliminated team would have no complaint about the fairness of the format, or at least a much less worthy complaint than a team eliminated in session 3 if the format was 5->3 on day 1, 3->2 in session 3, etc.

 

Also, I really dislike the idea of seeding an event like this in any fashion that requires the higher seed(s) to beat fewer teams to get to win the district (as opposed to having a slightly easier draw in a straight KO format, which seems sensible). It's tough enough to seed at the national KO level.

 

Actually answering this question would require some assumptions about relative team strength and then doing some calculations.

[gnasher]

Having the RR winner get to pick, combined with carry-over, allows targeting.

Yes. The idea was to give them more incentive to play well in the round robin, and to reward them for doing so. An alternative is to do a random draw for the semi-final, and rely on the carryover to encourage everyone to play well in the round robin.

 

What doesn't work is to use the round-robin places to determine the semi-final lineup by some formula - that would just provoke yet another Bridge World editorial.

Edited May 24, 2010 by gnasher