[Dean]

Hi

Does anyone have nay smart ideas for how to handle extreme shapes during symmetric relay auctions? We used to jump a round of bidding when showing controls to show this hand type but this becamse confusing very quickly.

For the one suiters i.e 8+ suits we can show them as 7+ and a void or 7+ and a singleton and treatign say 7+321 as 7331 for DCB's purposes but is there a better way?

Thanks
Dean

[rbforster]

I'm interested too. Right now there are a lot of "holes" in the (non-symmetric) relay structure I use. Bids higher than 2♠ are undefined for example, currently out of laziness and lack of a clearly good idea. Very high bids should be used for very specific hands, maybe something like 8221 or 8311 and a certain range of controls.

The kludge for our current method consists of showing a 7 card suit and accurate shortness, either 7330 when we've got 8320 for example, or 7321 when we have 8221. We can't show 2 shortnessness though, except by lying and "inventing" a second suit which seems wrong (ie calling 8311 or 9211 something like 7411). After that if we like our hand in context and if partner signs off at game (thinking we've got 7321), we might make a simple raise - i.e. 1♣...(showing 7♠321)...4♠ (signoff) - 5♠.

[Free]

Personally I wouldn't bother about these extreme hands. Most of them can be described as similar holdings (try not to lie about singletons and voids), for example an 8320 can easily be described as 7330. Usually opps intervene when you have an extreme hand anyway, so relays will be off most of the time.

The most extreme hand I've ever had playing a relay system (during several years) was AKQJTxxxx-xx-xx-void. Partner started with 1♣, opps were silent (damned for once). This time I didn't have a good way to tell my hand, because the auction started:
1♣ - 1♥
1♠ - 2♦
2♥ - ...
2♦ showed a singlesuiter with 6+♠. Since we show shortnesses from high to low, I would have to bid my shape immediately. The best description however is 7330, which is bid by 3NT (not forcing!). I could 'zoom' by bidding 4♣ or higher, but that would show at least 3 more AKQ-points, which I don't have (and could bring me in very serious trouble, like partner bidding 7NT)! I just blasted 7♠, heard partner laugh at the other side of the screen, but we made the slam when ♣A was outside, while partner had ♦AK and ♥AK.

This is the ONLY time that I had a real problem in more than 4 years time, so reserving some space for these extremely rare hands is a pure waste. Frequency vs usefulness must be in balance, so I wouldn't make my 'every day' relay schemes any worse for a 'once in 4 years'-hand.

[rbforster]

Free, on Sep 3 2007, 06:06 PM, said:

I just blasted 7♠

...reserving some space for these extremely rare hands is a pure waste.  Frequency vs usefulness must be in balance, so I wouldn't make my 'every day' relay schemes any worse for a 'once in 4 years'-hand.

Well apparently you didnt have a meaning for a 7♠ blast, so I guess it wasn't hurting your other bidding much . I would hazard most strong club systems haven't bothered to define very high jump bids, so if nothing else filling in the gaps with rare meanings couldn't hurt (although admittedly it won't show up very often).

More generally, with extreme single suiters, it might be worth turning captaincy over to responder to ask for aces and such. If you're sure you know what suit is trump (since you've got 8+ decent ones), it might be worth having a bid or bids where responder can ask for aces, then kings, maybe exclusion in case he has a void, etc. Depending on your preferences, this could happen before or after a first round transfer to the long suit. With such a distributional hand, it might be worth trying to shut out possible preemption by 4th hand by having the bidding start 1♣-3NT* or whatever you want your unusual asking bid to be.

[mikeh]

I have never played symmetric relay (so why am I posting? ) but I have played a fairly complete relay method. I echo the earlier advice about not worrying about 8 card suits.. treat it as a 7 card suit. Otherwise, you are simply getting too high and consuming your own bidding space. If this inaccuracy is costly, console yourself with the knowledge that the rest of the field/the opps are no better off than you are... no system handles freaks well, and I am willing to bet that no system will ever be invented that does: there is simply not enough bidding space. Treat 8320 as 7330, and hope partner doesn't insist on playing in your 2 card suit as trump, opposite his 5 cards...

[MarceldB]

mikeh, on Sep 5 2007, 01:50 AM, said:

... no system handles freaks well, and I am willing to bet that no system will ever be invented that does: there is simply not enough bidding space.

I'm sorry to disappoint you
As an extension of the Delta singleton system/scheme I can treat the 8-12carder as well, including a min or a max within the pointrange.

F.e. the hand Free gave:

▪ Pass=13+
2♥= 6-11p. singl./void ♣ and a 4441 or 5?440, OR 5-10p. and a 822=1 or 922=0 with a max or 10 11=1 with a min., 10 21=0 with a min, or finally a 11 1-1=0 with a max
▪2NT= relay
3♣= 4441 min or a 5=4=4=0 min/max, or 5-10p. and a 8=2=2=1 or 9=2=2=0 with a max or 10=111 min., 10=21=0 with a min, or finally a 11=1=1=0 with a max
▪3♦= relay
4♦= 9=2=2=0 and 9-10p.
▪4♥= Slam asking bid♠
5♠= AKQJ♠, no J♥ (loose Jacks do not count namely)
▪7♠

or if you prefer this spiral scan method I use:
▪4♥= control points (A,K only)
4NT= 3 CP
▪5♣= scan
5♦= non or AK♠
▪5♥= Queen♠?
6♣= yes + J♠, no J♥
7♠

In the framework of a freakhand the 4♦ bid gives the exact distribution + the HCP with a margin of 1 point only, so still well treatable in my opinion. Don't forget that Relayer did not break the relaysequence.
If a relay break with f.e. 3♥ after 2♥, RR still can give his exact distribution
with his 8+ carder ♦, ♥, or ♠, and on a safe level. Moreover with the same bids as in the relay sequence.

For more info: Deltex scheme

[helene_t]

In deWijs/Muller's "Tarzan" Precision (which is basically Wei with 2♣=6+ and 1♥ with 5/5 majors, i.e. very similar to the original SR), the openings are allocated such that all shapes without an 8-card and without 11 cards in the two longer suit can be resolved below 3NT sparing one step of bidding space* which (I think) is used for control count.

Presumably, the wilder hands will have to bypass 3NT but even then I don't think they can be resolved excactly. I would not bother abot it, the wilder your hand the bigger the odds that opps will mess up your relays anyway.

The problem only applies to the 1♠ and 2♣ openings. The 1♥ opening has plenty of steps left if you use 1♠ as the initial relay. The same would be true for 1♦ if they used 1♥ (or 1♠) as the initial relay, but they actually use 1NT so maybe they can't resolve the wild 1♦ hands.

*If this applies to the 2♣-opening as well is not clear to me. Rob Helle wrote a series about the system in IMP. Some hands with a six-card clubs would have to open something else (e.g. always 1♦ with 4-6 minors) in order to meet the 1-step-left-after-shape requirement, but whether they actually do so I don't know.

[hrothgar]

helene_t, on Sep 5 2007, 12:51 PM, said:

In deWijs/Muller's "Tarzan" Precision (which is basically Wei with 2♣=6+ and 1♥ with 5/5 majors, i.e. very similar to the original SR), the openings are allocated such that all shapes without an 8-card and without 11 cards in the two longer suit can be resolved below 3NT sparing one step of bidding space* which (I think) is used for control count.

Ick...

I have more than a passing familiarity with Symmetric Relay. In theory, you could reallocate some bidding space and allow some of the extreme shapes to be specifically resolved. In practice, the major impact will be displacing your auction termination structures much higher. Somethings got to go (and it is going to be that nice 3♥ control ask that we all love so much).

I can't recall more than a handfull of times where I had the opportunity to resolve an 8+ card suit. I agree with earlier comments to (essentially) ignore the problem and consider this as a cost of doing business

[Free]

mikeh, on Sep 5 2007, 12:50 AM, said:

no system handles freaks well, and I am willing to bet that no system will ever be invented that does: there is simply not enough bidding space.

Original MOSCITO-relays handle ALL shapes, but they consume more bidding space than symmetric on most hands.

(frequency * consumed bidding space) is as good as optimal for symmetric relay, which is never the case for relay schemes that can handle freaks. Although the freaks are out of the symmetric relays, I still consider it the best possible scheme for most hands.

[gwnn]

that deltex thingie is totally parallel with bridge principles. no 13-000???

[keylime]

In Ultra, we do not worry about any 8 card suit holdings - the maximum we have is a seven card holding which is high up the response schedule when it happens. We primarily focus on the 5431 and 55 shapes when we show an unbalanced hand.

[rbforster]

hrothgar, on Sep 5 2007, 05:30 AM, said:

In theory, you could reallocate some bidding space and allow some of the extreme shapes to be specifically resolved.  In practice, the major impact will be displacing your auction termination structures much higher.  Somethings got to go (and it is going to be that nice 3♥ control ask that we all love so much).

I understand the whole (frequency * level) issue where you trade off the level of the bid at which you resolve exact shape against the probability of such shapes showing up. That said, I don't think the costs to including extreme shapes would be very much, or rather, that they come at the expense of only-slightly-less-extreme shapes which weren't very likely either.

For example from symmetric relay (from here), your 1-suiter table resolves like this

2S high shortage
2NT middle shortage
3C equal shortage (7222 or 6322)
3D low shortage, zoom (5332)
3H low shortage, 6331
3S low shortage, 7321
3NT low shortage, 7330

Now a specific 7330 is only a little more common than a specific 8221 (by 4:3 or so), so it's not entirely unreasonable to continue the single-suiter table like this:

3N low shortage, 7330
4C low shortage, 8221
4D low shortage, 8320
4H low shortage, 9310, ...

What does this cost you? Typically after using a shape relay (symmetric or otherwise) to resolve exact shape, relayer will have the option of relaying again to ask about strength (controls, points, keycards, etc). Using the "zoom" principle, where you would start answering strength if you hit the last shape in the table, before our single suiter table really looked like this:

3S low shortage, 7321
3N low shortage, 7330, and minimum strength
4C low shortage, 7330, first step of extra strength
4D low shortage, 7330, second step of extra strength ...

hrothgar, on Sep 5 2007, 05:30 AM, said:

Somethings got to go (and it is going to be that nice 3♥ control ask that we all love so much).

So adding extra shapes will make it a little harder to resolve the strength of the 7330 hands (since you can't zoom), at the expense of being able to show more extreme shapes. This is a much more minor cost I think than Richard suggests - it's not like somehow our balanced hand relays lose their 3♥ strength ask since they finish higher than 3♦.

Of course at some point you'll run into the issue of resolving shapes too high. When you haven't promised enough values to make more than game, so you don't want your minimum 8221 hand pushing past 4♠ for example. But that aside, I don't see why everyone's so enamored of getting a little extra edge in their 7330 auctions but don't care about their 8221's.

[joshs]

Rob F, on Sep 5 2007, 01:43 PM, said:

hrothgar, on Sep 5 2007, 05:30 AM, said:

In theory, you could reallocate some bidding space and allow some of the extreme shapes to be specifically resolved.  In practice, the major impact will be displacing your auction termination structures much higher.  Somethings got to go (and it is going to be that nice 3♥ control ask that we all love so much).

I understand the whole (frequency * level) issue where you trade off the level of the bid at which you resolve exact shape against the probability of such shapes showing up. That said, I don't think the costs to including extreme shapes would be very much, or rather, that they come at the expense of only-slightly-less-extreme shapes which weren't very likely either.

For example from symmetric relay (from here), your 1-suiter table resolves like this

2S high shortage
2NT middle shortage
3C equal shortage (7222 or 6322)
3D low shortage, zoom (5332)
3H low shortage, 6331
3S low shortage, 7321
3NT low shortage, 7330

Now a specific 7330 is only a little more common than a specific 8221 (by 4:3 or so), so it's not entirely unreasonable to continue the single-suiter table like this:

3N low shortage, 7330
4C low shortage, 8221
4D low shortage, 8320
4H low shortage, 9310, ...

What does this cost you? Typically after using a shape relay (symmetric or otherwise) to resolve exact shape, relayer will have the option of relaying again to ask about strength (controls, points, keycards, etc). Using the "zoom" principle, where you would start answering strength if you hit the last shape in the table, before our single suiter table really looked like this:

3S low shortage, 7321
3N low shortage, 7330, and minimum strength
4C low shortage, 7330, first step of extra strength
4D low shortage, 7330, second step of extra strength ...

hrothgar, on Sep 5 2007, 05:30 AM, said:

Somethings got to go (and it is going to be that nice 3♥ control ask that we all love so much).

So adding extra shapes will make it a little harder to resolve the strength of the 7330 hands (since you can't zoom), at the expense of being able to show more extreme shapes. This is a much more minor cost I think than Richard suggests - it's not like somehow our balanced hand relays lose their 3♥ strength ask since they finish higher than 3♦.

Of course at some point you'll run into the issue of resolving shapes too high. When you haven't promised enough values to make more than game, so you don't want your minimum 8221 hand pushing past 4♠ for example. But that aside, I don't see why everyone's so enamored of getting a little extra edge in their 7330 auctions but don't care about their 8221's.

Losing the runons is not the only problem. You also lose the ability to play 4M in your 8 or 9 card suit which you would have gotten to do if you had merely showed 7 cards.

For instance in your scheme with 3820, the shape is resolved at the 4D level. You can no longer play 4H since the next step is a relay. You can decide to agree that 4S is the relay instead which
a. is an accident wiating to happen, since this auction occurs only once per year
b. This doesn't let you play 4S....

In general, shape resolution above 4C is a very bad idea (as is 3N showing shape but not strength).

In any case here are my estimates:
a. Being able to resolve unusual shpaes helps you once every 3 years
b. Losing space due to the runons hurts you 5 times/year

In general, when you have extreme shape, pretend you have an extra queen, find the closest approximation of your shape, and bid accordingly.

(For instance, my min positive range is normally 9-12, but with 5-5 or 64- it becomes 8-11, with an 8 card suit, or a very good looking 7420 its more like 7-10. I think Kxx AKxxxxxx - xx is a "maximum" so I would show 3712, and if asked, show a max. Note: In my version of symmetric relay you can't even distinguish 3712 from 2713....)

Summary: Bridge is about getting the common hands right. No one claims they are experts of how to bid wierd hands well, so don't worry about them.... And when you have an 9 card suit, how often are the opps silent????

[awm]

One thing that helps in Sam and my methods is the option of reversing the relays. We play a multi-way 1♦ response to 1♣ that's any of:

(1) A very strong hand, making other direct responses limited to 6 AKQ points.
(2) A balanced game force.
(3) A very weak hand.
(4) A game forcing hand that prefers to relay rather than describe.

This last option takes care of a lot of the 8-card suits etc.

[Cascade]

Curiously in my symmetric relay partnership we are going in the other direction.

We have recently moved 7411 into extreme shape and do not resolve this shape exactly.

[hrothgar]

Rob F, on Sep 5 2007, 09:43 PM, said:

hrothgar, on Sep 5 2007, 05:30 AM, said:

In theory, you could reallocate some bidding space and allow some of the extreme shapes to be specifically resolved.  In practice, the major impact will be displacing your auction termination structures much higher.  Somethings got to go (and it is going to be that nice 3♥ control ask that we all love so much).

I understand the whole (frequency * level) issue where you trade off the level of the bid at which you resolve exact shape against the probability of such shapes showing up. That said, I don't think the costs to including extreme shapes would be very much, or rather, that they come at the expense of only-slightly-less-extreme shapes which weren't very likely either.

For example from symmetric relay (from here), your 1-suiter table resolves like this

2S high shortage
2NT middle shortage
3C equal shortage (7222 or 6322)
3D low shortage, zoom (5332)
3H low shortage, 6331
3S low shortage, 7321
3NT low shortage, 7330

Now a specific 7330 is only a little more common than a specific 8221 (by 4:3 or so), so it's not entirely unreasonable to continue the single-suiter table like this:

3N low shortage, 7330
4C low shortage, 8221
4D low shortage, 8320
4H low shortage, 9310, ...

What does this cost you?

Given how few people here play much relay, its surprising to see this topic generate much in the way of discussion. Regardless, here are a few points that folks might want to consider:

I just ran a quick simulation. It suggested that the relative frequency of an 7330 pattern compared to an 8220 was skewed much more than 4:3. Does anyone have a copy of the Encyclopdia of Bridge handy (Mine is packed away from the latest move)

First, here's a frequency distribution for the 7330 patterns as well as a few of the "extreme" single suited patterns. (Please note, I only generated a million hands so these should be viewed as ballpark figures)

Shape = 7330, Frequency = .002088
Shape = 8221, Frequency = .000158
Shape = 8320 / 8230, Frequency = .000082
Shape = 9310 / 9130, Frequency = .0000068
Shape = 9220, Frequency = .000005

We're talking about VERY small numbers here.

Second: When you're looking at some of the hand patterns, you know the long suit and the short suit. You don't have any information about the residuals. This could be important, especially if you are dealing with 9310 patterns.

Third: If folks really wanted to, I suppose that you could zoom with extreme shape and single suited hand patterns. (I wouldn't want to do so with two suiters where you might reasonably want to play in one of two strains). If you were going to do this, you'd probably want to set a rule that the this zoom automatically sets responder's long suit as trump. In turn, this would let you use standard auction termination rules. (This neatly sidesteps AWM's issues with memroy loads with respect to 4♥ / 4♠ bids and the like)

[Cascade]

hrothgar, on Sep 6 2007, 10:19 AM, said:

I just ran a quick simulation. It suggested that the relative frequency of an 7330 pattern compared to an 8220 was skewed much more than 4:3. Does anyone have a copy of the Encyclopdia of Bridge handy (Mine is packed away from the latest move)

First, here's a frequency distribution for the 7330 patterns as well as a few of the "extreme" single suited patterns. (Please note, I only generated a million hands so these should be viewed as ballpark figures)

Shape = 7330, Frequency = .002088
Shape = 8221, Frequency = .000158
Shape = 8320 / 8230, Frequency = .000082
Shape = 9310 / 9130, Frequency = .0000068
Shape = 9220, Frequency = .000005

We're talking about VERY small numbers here.

7330 0.002652452387097
8221 0.001923576007998
8320 0.001085094158358
9310 0.000100471681329
9220 0.000082204102906

[MarceldB]

gwnn, on Sep 5 2007, 06:29 PM, said:

that deltex thingie is totally parallel with bridge principles. no 13-000??? 

I understand your understatement, but I hate it to preëmpt my partner, even with a 12-100 I still build it up.

W/non

1♣ - 1♦
1NT - 2♣
2♥ - 2♠
2NT - 3♣
4♠ - 5♥
Pass

---

1♣ = 8-12 any s/void (with a 12 100 then HCP do not count)
▪ 1♦ = relay
1NT = s/v ♦ , L= Major or any 5/5
▪2♣ = relay
2♥ = L= ♥ + 4 card else
▪2♠ = relay
2NT= 4c ♣
▪3♣ = relay (or do a positive break with 3♦ = stopper asking ♠ + ♣)
in both cases: 4♠ = 0=12=0=1 and a minimum = 11 tricks
5♥ = final bid

If East is the dealer:
Pass - 1NT= 6-11 s/v ♦ , L= Major or 5/5
etc. and the same result
Turn the minor holding in East and you can claim 6♥

Happy with a 5♥ opening? Or can somebody ask for Ace ♥ and/or ♣ ?
And how to handle this hand if East opens 1♠ ?
===
Such occasions for above mentioned are rare and specially in case of no interference, but in the framework of the system it was no problem at all to build-in those extreme 1-suiter shapes without disturbing the regular shapes AND without getting too high, so I have the opinion: why not.

Regards,
Marcel

[helene_t]

hrothgar, on Sep 6 2007, 12:19 AM, said:

Given how few people here play much relay, its surprising to see this topic generate much in the way of discussion.

I get your point, but as an applied mathematician I am specialized in forming opinions about issues that I don't know much about, so I'm excused.

[gwnn]

A 12-1-0-0 comes up every 8578 years if you play 100 hands a day. But I think I messed up the calculations somewhere, it really has to be less often. I don't mean this to criticize your system mate, don't get me wrong.