Do you relay at the 6-level?

[pilun]

[hv=pc=n&w=sjhaktdakj2ckj853&e=sakt9873h964dq6ca&d=w&v=b&b=4&a=1cp1h(4%2BS)p1sp3s(7-%5B3%2B2%5D-1)p4cp5c(9%20SPs)p5dp5h(4%2B%20SPs%20in%20S)p5sp5n(No%20H%20honour)p]266|200[/hv]

 

So 9 SPs and no ♥Q means East has to have ♠AK & ♣A plus either ♠Q or ♦Q.

 

Of course ♠Q is 3½ times as likely as ♦Q, so you might just go with the odds if 6♣ is not a relay.

 

Best is to relay with 6♣ if available, stopping in 6♠ opposite the actual.

 

Our long-standing agreement is that 5NT is the last DCB-ask, maybe from an unpleasant overstep decades ago. I forget.

 

What is your agreement and do you think it's clear?

[pilun]

Two questions for the price of one

 

How do you deal with known singletons in DCB?

 

For decades, we have switched to positive cueing with those, so bidding the singleton step = stiff A or K (stiff king counted as 1 SP, ignore stiff queens)

 

I'm told the gain for this approach is very rare, though better than standard DCB. I believe (?) that most pairs just ignore the singleton suit, depending on DCB in the other three to sort it out.

 

Is that right?

[awm]

We play relays at the six level (but not 6NT or above), so 6♣ would be a relay.

 

Our honour locations are based on parity, so after 5♦ relay we would have:

 

5♥ = 0 or 2 of ♠AKQ

 

This is already enough to know that there is no grand; if we did relay with 5♠ and get 5NT (0 or 2 of ♥AKQ) we would know the entire honour structure of the hand. At least in this case, we are a bit ahead of whatever you are doing with 4+SP.

 

When it comes up, we don't do anything special with singletons (still parity, stop=no honor and skip=stiff honor).

[foobar]

Two questions for the price of one

 

How do you deal with known singletons in DCB?

 

For decades, we have switched to positive cueing with those, so bidding the singleton step = stiff A or K (stiff king counted as 1 SP, ignore stiff queens)

 

I'm told the gain for this approach is very rare, though better than standard DCB. I believe (?) that most pairs just ignore the singleton suit, depending on DCB in the other three to sort it out.

 

Is that right?

 

I have played various agreements, including counting stiff K=1 and even stiff A=2 (FWIW I really disliked it, but pard insisted). On the other end of the spectrum, we have agreements like awm's approach that discounts only stiff Qs (which might actually be better).

 

For the stiff A/K, I thought that the "standard" DCB method was to invert, and stop with, and skip without. Even if it isn't standard, it might be more efficient, depending on what happens after the first scan.

 

Regarding the highest DCB, mostly used awm's method of 6N as not being a relay).

[nullve]

https://www.bridgebase.com/forums/topic/79072-do-you-relay-at-the-6-level/

[straube]

I have played various agreements, including counting stiff K=1 and even stiff A=2 (FWIW I really disliked it, but pard insisted). On the other end of the spectrum, we have agreements like awm's approach that discounts only stiff Qs (which might actually be better).

 

https://www.bridgebase.com/forums/topic/56295-stiff-a-3-qps/

 

I'm taking credit for Q=0 while scanning it as an honor. Counting stiff ace as 2, stiff K as 1 and Q as 0 helps judge when it is safe to search for slam because (as others have described) stiff honors are worth roughly a queen point less.

[foobar]

https://www.bridgebase.com/forums/topic/56295-stiff-a-3-qps/

 

I'm taking credit for Q=0 while scanning it as an honor. Counting stiff ace as 2, stiff K as 1 and Q as 0 helps judge when it is safe to search for slam because (as others have described) stiff honors are worth roughly a queen point less.

Since awm doesn't discount stiff Ks, surely this idea isn't mainstream. Adam, what has your experience in this matter been?

[awm]

Since awm doesn't discount stiff Ks, surely this idea isn't mainstream. Adam, what has your experience in this matter been?

 

I commented on the other thread, and my opinion hasn't changed. I don't like it, for the following reasons:

 

1. It makes it more difficult to use shortcuts like subtracting the RP count from 24. This might not matter given unlimited time to think about the possible hands, but in practice it does make a difference.

2. While this will help you a bit on hands where a singleton honor is not useful, it will hurt you a LOT on hands where that singleton honor matters. Overall I don't think that's a good trade.

3. Things like bidding RKC in the short suit aren't a solution -- typically this suit is LAST in the ordering of RKC bids, and there are also times when this is relayers second suit and he doesn't really want it as trumps.

[straube]

This method disambiguates singleton honors including the queen. Certainly often a singleton honor is exactly what is wanted, but I think most of the time a singleton honor is a disappointment; that's my own experience anyway. The evaluation 0, 1, 2 is not perfect but is a better approximation most of the time.

 

Apparently Atul forgot that you didn't discount the stiff queen (which is why he thought this might be better).

[foobar]

This method disambiguates singleton honors including the queen. Certainly often a singleton honor is exactly what is wanted, but I think most of the time a singleton honor is a disappointment; that's my own experience anyway. The evaluation 0, 1, 2 is not perfect but is a better approximation most of the time.

 

Apparently Atul forgot that you didn't discount the stiff queen (which is why he thought this might be better).

 

FWIW, I am OK with discounting the stiff Q, ambivalent about discounting stiff K, and think that discounting stiff A is stretching things too far.

[pilun]

I looked at a few resources.

 

Truscott (= Rigal)

Relays continue at the 6-level. 6NT+ are to play.

They seem to count singleton honours normally (not sure) but ignore singletons in DCB. That saves steps at the expense of some ambiguity.

 

Say describer has some 5431 and the bid that says "Now look at the third suit" was 5♣. Assuming a small singleton, we get

 

Truscott

5♣ - 5♦ = a second look at the 5-carder (so -2)

Abraham and us

5♣ - 5♥ = assumed small singleton, then looking a the 5-carder (so -1)

AWM (?)

5♣ - 5♦ = small singleton - 5♥ - 5♠ = second look at the 5-carder (so square)

 

There are costs in ignoring singletons.

♦AKQx ♣x might be treated the same way as

♦Axxx ♣A, or even

♦KQxx ♣A

 

finding out which might not be practical.

 

For singletons, we count A=3, K=1, Q=0 and have done so for decades. Resolving helped once when describer had either

♦Kxx ♣K, or

♦Axx ♣x

 

when looking at the singleton was crucial. Of course, treating kingletons as 2 would have avoided that. At a cost.

 

Given that ♠Axxxx ♥KQxx ♦xxx ♣x is likely to be more useful than

♠Axxxx ♥Qxxx ♦xxx ♣K

 

it's good that they are evaluated differently. The danger comes when asker pictures the second hand as a more typical 5 SP hand

 

♠Axxxx ♥xxxx ♦Kxx ♣x

 

though experience suggests that these tend to come out okay.

 

Kit only counts controls. His DCB also ignores singletons. In fact he ignores the THIRD suit when there is a singleton in the 4th, preferring to go back to the two long suits before having a first look at the 3rd suit. Seems a bit extreme.

[straube]

knr for xxxx xxxx xxxx A = 4.4

knr for Axxx xxxx xxxx x = 5.6

knr for Kxxx xxxx xxxx x = 4.6

 

I chose 4441s because the average suit length for a hand with a stiff is obviously 4.

 

So far the knr would evaluate the stiff A more like a king from length. On many days, that king would be supported by other honors or minor honors and would be worth even more, so at least as far as knr, I think a stiff ace is worth approximately a non-stiff king.

 

On most days I'll want to find AQxxx Kxxx xxx x instead of Qxxxx Kxxx xxx A, but if I count a stiff ace as 2 and hope for AQxxx Kxxx xxx x (knr 11.2), I might survive with Qxxxx KQxx xxx A (knr 11.15) or I might be disappointed to learn that I might have two spade losers if I hold Kxx (when I would have preferred to lose one club).

 

Some side notes. Isn't the SQ roughly 2.5 times as likely as the DQ since the SA and SK are known? Also, I followed awm's recommendation for relaying at the 6-level (6N being always to play) and found it more useful to relay than to sign off.

[pilun]

knr for xxxx xxxx xxxx A = 4.4

knr for Axxx xxxx xxxx x = 5.6

knr for Kxxx xxxx xxxx x = 4.6

 

I chose 4441s because the average suit length for a hand with a stiff is obviously 4.

 

So far the knr would evaluate the stiff A more like a king from length. On many days, that king would be supported by other honors or minor honors and would be worth even more, so at least as far as knr, I think a stiff ace is worth approximately a non-stiff king.

 

On most days I'll want to find AQxxx Kxxx xxx x instead of Qxxxx Kxxx xxx A, but if I count a stiff ace as 2 and hope for AQxxx Kxxx xxx x (knr 11.2), I might survive with Qxxxx KQxx xxx A (knr 11.15) or I might be disappointed to learn that I might have two spade losers if I hold Kxx (when I would have preferred to lose one club).

 

Some side notes. Isn't the SQ roughly 2.5 times as likely as the DQ since the SA and SK are known? Also, I followed awm's recommendation for relaying at the 6-level (6N being always to play) and found it more useful to relay than to sign off.

 

Yes, 2½ times as likely to be ♠Q.

 

The issue with stiff ace is not so much that it's worth less, rather whether relayer has a chance to diagnose it.

♠xxxxx ♥xxxx ♦Axx ♣A is still worth a positive.

 

Thanks for another vote in favour of relaying at the 6-level, though not sure whether relaying with 6♠ over 6♥ will be welcome.

[straube]

Yes, 2½ times as likely to be ♠Q.

 

The issue with stiff ace is not so much that it's worth less, rather whether relayer has a chance to diagnose it.

♠xxxxx ♥xxxx ♦Axx ♣A is still worth a positive.

 

Thanks for another vote in favour of relaying at the 6-level, though not sure whether relaying with 6♠ over 6♥ will be welcome.

 

I see. I use IMprecision responses so get to describe that as a semipositive/light GF and then force game. So to it's slightly less awkward than presenting that as a negative.

 

There are many 5 QP hands worth a GF; have you thought about grouping 5-6 QPs into your step one QP response and then separating them later if asked? Often opener won't even be interested. Then 5 QP hands could GF optionally.

[pilun]

I see. I use IMprecision responses so get to describe that as a semipositive/light GF and then force game. So to it's slightly less awkward than presenting that as a negative.

 

There are many 5 QP hands worth a GF; have you thought about grouping 5-6 QPs into your step one QP response and then separating them later if asked? Often opener won't even be interested. Then 5 QP hands could GF optionally.

 

We group 4 & 5 but leave opener to guess, don't split them. So

 

♠KJxx ♥KJxxxx ♦xx ♣x

 

is a positive, for better or worse.

[straube]

I imagine 5 is much more common than 4, so assuming 5 is likely to work out most of the time. But then you could count Axxx xxxx xxxx A as GF and 5 if you wanted to. One other reason for deducting a point for a stiff honor (including the ace)is that the stiff is scanned last (if you scan it at all) which sometimes gives a late and unwelcome surprise.

[foobar]

I imagine 5 is much more common than 4, so assuming 5 is likely to work out most of the time. But then you could count Axxx xxxx xxxx A as GF and 5 if you wanted to. One other reason for deducting a point for a stiff honor (including the ace)is that the stiff is scanned last (if you scan it at all) which sometimes gives a late and unwelcome surprise.

In a scheme that groups 4 and 5, won't counting stiff A as 2 result in an unworkable 4-6 for the first step?

 

In other words, in Nick's example, we can have both ♠KJxx ♥KJxxxx ♦xx ♣x OR ♠QJxx ♥KJxxxx ♦xx ♣A with the proposed scheme that devalues stiff As.

[straube]

In a scheme that groups 4 and 5, won't counting stiff A as 2 result in an unworkable 4-6 for the first step?

 

In other words, in Nick's example, we can have both ♠KJxx ♥KJxxxx ♦xx ♣x OR ♠QJxx ♥KJxxxx ♦xx ♣A with the proposed scheme that devalues stiff As.

 

 

knr for KJxx KJxxxx xx x 10.5

 

knr for QJxx KJxxxx xx A 12.25

 

Seems reasonable to me to group those in a 4-5 QP range, both GF hands. Assume you disagree. If instead the A is counted as 3, we might hope for

 

QJxx AKJxxx xx x 14.5 when we have instead QJxx KJxxxx xx A 12.25, both counted here as 6 QPs.

[foobar]

knr for KJxx KJxxxx xx x 10.5

 

knr for QJxx KJxxxx xx A 12.25

 

Seems reasonable to me to group those in a 4-5 QP range, both GF hands. Assume you disagree. If instead the A is counted as 3, we might hope for

 

QJxx AKJxxx xx x 14.5 when we have instead QJxx KJxxxx xx A 12.25, both counted here as 6 QPs.

I think that as awm and others have noted, using KNR evaluation alone to justify devaluating stiff Aces is a very narrow perspective because it doesn't look at the combined potential. As awm noted, given an arbitrary amount of time to evaluate all possible hand combinations, such a method might be better, but on the flip side, it makes it more difficult to subtract RPs from 24 to figure out the level of the final contract. Specifically, for every gain there almost certainly has to be a hand where the presumed (incorrect) lack of RPs keeps us out of slam. Until there is compelling evidence that this method is superior, count me as a skeptic.

[pilun]

Related question

 

If you demote singleton honours (stiff K => 1, stiff Q => 0) do you do the same with dry honours? So

 

AK- tight. Does that become 4 SPs?

 

AQ - tight. Is that 3 SPs and ignore the queen entirely in DCB?

[straube]

Related question

 

If you demote singleton honours (stiff K => 1, stiff Q => 0) do you do the same with dry honours? So

 

AK- tight. Does that become 4 SPs?

 

AQ - tight. Is that 3 SPs and ignore the queen entirely in DCB?

 

I hadn't thought of that. Interesting idea though. Certainly HH is unexpected and blocks the suit, etc.

 

I use Parity Cue Bidding which I learned from Adam and Sieong. When pd was willing to deduct for stiff honors I felt my slam bidding improved.

 

I think Foobar and Adam deduct from 24 to see what could be missing. I've always just looked at how many QPs partner showed and then saw if he could have a combination of cards I needed for slam. So deducting QPs (or adding them conceivably) for certain honors never gave me any trouble.

[Gilithin]

I hadn't thought of that. Interesting idea though. Certainly HH is unexpected and blocks the suit, etc.

If you have QJT9xxxx in a suit, is it really bad news to find out that partner's doubleton is AK for slam purposes? Under what circumstances do you think this is going to be an advantage? Compare this with the question of whether partner's singleton suit has 0 or 1 loser when they have a singleton ace as opposed to a singleton king. Think about pros vs cons and you will surely come to the conclusion that downgrading doubleton honours makes little sense when it comes to slam scanning.

[straube]

If you have QJT9xxxx in a suit, is it really bad news to find out that partner's doubleton is AK for slam purposes? Under what circumstances do you think this is going to be an advantage? Compare this with the question of whether partner's singleton suit has 0 or 1 loser when they have a singleton ace as opposed to a singleton king. Think about pros vs cons and you will surely come to the conclusion that downgrading doubleton honours makes little sense when it comes to slam scanning.

 

Easy there. I just said I hadn't thought about it but it was an interesting idea (to me) and doubleton honors are usually worth less than accompanied by length. Not for or against deducting for them.

[DinDIP]

Re discounting doubleton honours: an Australian international asked me about this (in the context of a pseudo-relay system). I tested a very small sample of deals and concluded that it was not a good idea for AK or AQ but might be worth considering for KQ. As Adam has noted, the value depends in part on whether the partnership uses DCB (when it takes two scans to distinguish between Ax and KQ) and PCB (when this is known in one scan).

[Cascade]

[hv=pc=n&w=sjhaktdakj2ckj853&e=sakt9873h964dq6ca&d=w&v=b&b=4&a=1cp1h(4%2BS)p1sp3s(7-%5B3%2B2%5D-1)p4cp5c(9%20SPs)p5dp5h(4%2B%20SPs%20in%20S)p5sp5n(No%20H%20honour)p]266|200[/hv]

 

...

 

Of course ♠Q is 3½ times as likely as ♦Q, so you might just go with the odds if 6♣ is not a relay.

 

 

Isn't the SQ roughly 2.5 times as likely as the DQ since the SA and SK are known?

 

 

Yes, 2½ times as likely to be ♠Q.

 

We are comparing the chances of ♠AKQxxxx ♦xx and ♠AKxxxxx ♦Qx. The opponents have five spades and seven diamonds. Partner has five unknown spades and two unknown diamonds.

 

The number of combinations of ♠AKQxxxx is nine (small spades) choose four which is (9*8*7*6)/(1*2*3*4).

 

The number of combinations of ♦xx is eight (small diamonds) choose two which is (8*7)/(1*2)=28.

 

The number of combinations of ♠AKxxxxx is nine (small spades) choose five which is (9*8*7*6*5)/(1*2*3*4*5). That is precisely the same number has for ♠AKQxxxx.

 

The number of combinations of ♦Qx is eight (small diamonds) choose one which is 8.

 

So there are 28 combinations of ♠AKQxxxx ♦xx for every 8 combinations of ♠AKxxxxx ♦Qx. The former is precisely 3.5 times more frequent than the second.