[Cascade]

kenrexford, on Aug 19 2009, 03:46 PM, said:

Cascade, on Aug 18 2009, 07:54 PM, said:

I tried to estimate this.

I gave one hand a specific game going 5=1=4=3 opposite any Balanced hand.

The standard deviation for tricks in spades was 0.955707068

Second i started with a specific 3=4=3=3 opposite a random 5=1=4=3.

The standard deviation was then 0.829442584

Not surprisingly lower since 5=1=4=3 is more specific than any Balanced hand.

Next I looked at 5 spades and 1 heart and any minor distribution (which happens to mimic our system).

The standard deviation was 0.906209689 - still lower than the distributional hand judging.

Surprisingly I generalized a bit more to 5+ spades and 0-1 heart (0 only with exactly five spades) and the standard deviation was 0.864081015.

While the responder most certainly needs to be the captain for the initial decision - part-score, invite game, game, invite slam, slam.  It seems that most definitely that opener is better placed to make the final decision.

I also tried crunching some numbers on this. I took the standard deviation ratio of the inverse of the par 0.50000, multiplied by the ratio of the number of cards in spades in the average deal where 1♠ is the opening, divided by the number of side suits expected to split 2-3 or 3-2 because of a secondary, but unknown, fit.

Using this figure, I then extrapolated the likelihood of a stiff emerging against each of the three top honors, and then combinations of the same, with an averaged deviation expectancy for the unknown values of lesser honors (using 8, 9, 10, and J at a surrogate value of .25, .33, .50, and 1 respectively), plus or minus from an average expectation of 3.33 (the average length of a suit when esponder has 4-3-3-3 and 3-card spade support) times the average spread of these cards that is expected (each suit having honors, or no honors, respectively).

I then cross-checked these figures to corroborate the valuations, discerning that ties to higher honors were less likely than originally inspected, in a sense because the high honors take a "space" in the grid of calculated occurrences.

This final figure was then tweaked by 0.125% off base, to gain a more realistic view of what is actually happening.

I then did the same process in reverse, to account for Opener-directed inquiry, as opposed to Responder-directed inquiry, alternatively Opener-show versus Responder-show (a term I think fits). Comparing these two results yielded interesting insights.

Primarily, I determined that Winston has no clue how to bid bridge hands and that Cascade is out of his mind.

I hope that doesn't mean i am in your mind.

[Free]

Winstonm, on Aug 19 2009, 12:21 AM, said:

Quote

Your statement has nothing to do with what I said. Your statement about captaincy after a 1NT opener is "true", but mine is true in a general context. You assume I'm talking purely about a 1NT opener, while in fact my statement is also true after strong ♣ auctions and alike where you have 1 or 2 unlimited hands.

This all semantics. I said the same thing - unless responder invites opener back into the discussion. It still isn't captaincy. And in your examples you are not showing shape (which you argue is the ideal) but simply getting help chosing the best game.

Sorry, but I distinctly remember you using the phrase "balanced hand". NT hands a balanced hands. This whole discussion began with another person's claim that it is always better for the distributional hand to describe his shape to the balanced hand.

Again, I say nonsense - in standard bidding, anyway.

Quote

Why is yours only "true"? Because most of the time, captaincy is in fact transfered back to opener. This is so for invitational hands, but also in choice of games after a transfer, or after stayman. 1NT-2♦-2♥-3NT asks opener to pass or bid 4♥. 1NT-2♣-2♥-3NT asks opener to pass or bid 4♠. Etc.

Quote

The argumentation disappeared from the quote, but can you tell me what's wrong with that before calling common sense + experience with relays "rubbish"?

Because this thread dealt with standard bidding and not relay methods.

Quote

Just to add another argument:
Suppose you have a 5431 and partner describes his hand as a 2434. How can you evaluate your hand in a borderline case? If partner has ♣Axxx or xxxx it's great, if he has KQJx, KQxx or KJxx it's awful. On the other hand, if you described your hand as a 5431, partner will know what to do

.

Inviting partner to help make a choice is not the same as saying I must tell my partner my shape and he makes the complete decision.

Quote

And another:
Why do you think NT systems concentrate more and more on showing responder's hand? I'm talking about transfer extensions, puppets, 4-card transfers, splinters,... It's easy to have full relays after a 1NT opening, not even having to give up on transfers, so why isn't anyone doing this at top level?

Because the hand type is known with the nt bid.

Enough of this discussion. It makes no sense unless you are strongly seeking to implement relay structures into all sequences and trying to argue that this method is superior.

I remember you saying the following:

Quote

it is certainly not true that it is better for responder to describe his hand to an opening NT bidder

I've given you arguments against this statement, and all you can throw back at me is "I don't know why, but it's rubish, nonsense,...", avoiding any kind of dialogue, and obviously followed by "I'm getting out of this discussion as quick as possible" when you start realizing you're wrong.

You haven't given any counter examples or arguments to convince anyone you're right. You just throw in a statement which you can't defend and when someone points out to you that it's wrong, you're gone.

Since I won't be able to convince you that your above statement is wrong, perhaps Ken's numbers may. One thing I know for sure is that his conclusion is correct.

[helene_t]

I like Cascade's research.

It occurs to me that Winston is discussing some different issue than Free and Cascade, not sure which though.

[Cascade]

Thanks Helene

To be really meaningful I would have to do a lot more samples. I only investigated one specific hand in each case.

[Winstonm]

helene_t, on Aug 19 2009, 09:01 AM, said:

I like Cascade's research.

It occurs to me that Winston is discussing some different issue than Free and Cascade, not sure which though.

I find that happens a lot around here - which is why I don't want to get drawn into these debates. I never know the rules.


I certainly agree that a case can be made that relay structures with the distributional hand describing his shape to a balanced hand can at times be superior in some ways.

However, you need to define the argument to make the case. The better methods in my view are cooperative bidding approaches where both players come to an agreement on the maximum place to play.

[Winstonm]

This was your original comment about this subject, Free.

Quote

It's general knowledge that in case of a balanced hand vs an unbalanced hand, that it's always better to describe the unbalanced hand.

That statement is hogwash because of the phrase "always better".

AKQx
87
KJ10
K109x

x
J10964
Qxxxx
Qx

My choice is to play this hand in 1NT, not to have the weak distributional hand describe its shape.

It is always hogwash because you are refusing to address the considerable problems inherent when 2N is not available as a natural and forcing bid.

No bid or systemic choice is made within a vacuum. If you elect to use 2N as a forcing, balanced raise, in order to have opener describe his hand then you must compensate for those balanced forcing hands that do not hold support.

It's like saying it's "always better" to fill you car's gas tank to the top - well, that may sound correct but it doesn't take into account whether or not you have the money to pay for the gas.

At least Ken made sense by saying "generally better". That may or may not be true, IMO - it depends on perspective.

[Free]

Winstonm, on Aug 20 2009, 09:33 PM, said:

This was your original comment about this subject, Free.

Quote

It's general knowledge that in case of a balanced hand vs an unbalanced hand, that it's always better to describe the unbalanced hand.

That statement is hogwash because of the phrase "always better".

AKQx
87
KJ10
K109x

x
J10964
Qxxxx
Qx

My choice is to play this hand in 1NT, not to have the weak distributional hand describe its shape.

It is always hogwash because you are refusing to address the considerable problems inherent when 2N is not available as a natural and forcing bid.

No bid or systemic choice is made within a vacuum. If you elect to use 2N as a forcing, balanced raise, in order to have opener describe his hand then you must compensate for those balanced forcing hands that do not hold support.

It's like saying it's "always better" to fill you car's gas tank to the top - well, that may sound correct but it doesn't take into account whether or not you have the money to pay for the gas.

At least Ken made sense by saying "generally better". That may or may not be true, IMO - it depends on perspective.

Well ok, "allways" is perhaps exaggerated. But your example, come on, can't you do better than this? Passing 1NT with the South hand is imo unrealistic...

[helene_t]

Winston, I think you are discussing something different.

Nobody is claiming that every bidding system will benefit from introducing a range of gadgets that allow the balanced hand to be captain in all situations. There are many situations where such gadgets would have disadvantages. For example, in a 5-card major system, an opening in a major suit is often unbalanced, yet making opener captain has the advantage of revealing less of declarer's hand in the likely event that we end up playing in opener's suit.

Also, nobody is claiming that some hypothetical bidding system that allows the balanced hand to always be captain would per saldo be superior to every other bidding system.

The issue is rather different. Cascade's research illustrates the point: That other things being equal, the balanced hand has a better chance of guessing the right contract based on information about partner's shape than the converse. I suppose this is largely because the unbalanced hand, by announcing shortness in a particular suit, allows the balanced hand to assess:
- Are my values in that suit likely to be wasted?
- Is my holding in that suit strong enough to play in notrumps?
Of course, the unbalanced hand could also make the decision, given a description of the balanced hand. But that would require more information exchange. The balanced hand would have to announce, among other things, in which suits it has wastage opposite a singleton, and in which suits it has sufficient stoppers opposite a singleton.

[Cascade]

helene_t, on Aug 21 2009, 09:31 PM, said:

Winston, I think you are discussing something different.

Nobody is claiming that every bidding system will benefit from introducing a range of gadgets that allow the balanced hand to be captain in all situations. There are many situations where such gadgets would have disadvantages. For example, in a 5-card major system, an opening in a major suit is often unbalanced, yet making opener captain has the advantage of revealing less of declarer's hand in the likely event that we end up playing in opener's suit.

Also, nobody is claiming that some hypothetical bidding system that allows the balanced hand to always be captain would per saldo be superior to every other bidding system.

The issue is rather different. Cascade's research illustrates the point: That other things being equal, the balanced hand has a better chance of guessing the right contract based on information about partner's shape than the converse. I suppose this is largely because the unbalanced hand, by announcing shortness in a particular suit, allows the balanced hand to assess:
- Are my values in that suit likely to be wasted?
- Is my holding in that suit strong enough to play in notrumps?
Of course, the unbalanced hand could also make the decision, given a description of the balanced hand. But that would require more information exchange. The balanced hand would have to announce, among other things, in which suits it has wastage opposite a singleton, and in which suits it has sufficient stoppers opposite a singleton.

I suspect this would be possible and perhaps just as efficient as showing distribution from the other side. But our traditional methods do not deliver this information.

[helene_t]

lol, now I am getting confused, maybe Winston is the one who understood what the issue was and I was wrong. Reminds me of a great post:

Winstonm said:

The purpose of this thread is to insult morons who think they know what the purpose of the thread is. If you want a thread about optimal agreements, start your own.

Anyway, since there are more unbalanced distributions than balanced ones, the balanced hand would need more information to know his partner's exact distribution. But consider this hypothetical bidding system:
- Balanced hands start with a forcing pass, then relay out partner's appr. strength and exact shape.
- Unbalanced hands open and allow partner to relay out their appr. strength and exact shape.

One could compare that to the reverse, and I would predict that the first system would work better.

[kenrexford]

There are two major goals in any slam-ward auction, which is the relevant auction once GF has been established.

1. Distributional Fit (what's trumps? What ruffing values or long suit values? Etc.)
2. Honor fit (what do these honors do for you?)

The balanced hand describing his balanced hand serves one function -- what is trumps.

The unbalanced hand describing his unbalanced hand serves multiple functions as to the distributional fit. What is trumps? What trick sources exist? What ruffing values exist?

The question then is how to assess honor fit. There are two ways to do this.

First, one could describe honors in the abstract:

1. Ace only
2. King only
3. Queen only
4. AK
5. AQ
6. KQ
7. AKQ
8. none

Add in Jacks and 10's and you exponentially increase the problem.

However, when distribution is known, the table changes, IMO:

1. We have a fit here OR a side fit here:
a. one top honor
b. two top honors
c. three top honors
d. four top honors

2. Someone has a doubleton here (or this is a control-only suit):
a. Ace or King
b. Ace and King
c. No Ace or King

3. Someone has a stiff here:
a. Ace
b. no Ace

4. Someone has a void here
a. Ace
b. No Ace

This is a simplistic model, but I think it illustrates the point. In the abstract, there are at least 8 permutations of possible interest. When a distributional feature is known, the number of permutations of interest reduces by at least half, if not more.

Now, Precision asking bids typically look at things this way, as the various types of asking bids are focused on the type of suit being discussed, thereby limiting the probe. However, this type of asking bid structure requires a low start, like sometimes 1-level for the first asking bid, because of the space needed to resolve things. That often is not enough room even then. In a 2/1 sequence, you generally start at the level of 2♠ as the lowest possible asking bid, meaning that there is very little space to work this out with an asking-bid structure. Make the call 2NT without fit resolved yet or 3M for that start, and how can you do more than ask one question about one suit.

[Cascade]

kenrexford, on Aug 22 2009, 12:36 AM, said:

The balanced hand describing his balanced hand serves one function -- what is trumps.

I think it would be possible for the balanced hand to describe more than this. Schemes for showing that information are not commonly played.

But it ought to be possible to show information like: I have a weak doubleton; I have a hand rich in controls etc. Whatever was felt important.

[Winstonm]

Quote

Well ok, "allways" is perhaps exaggerated. But your example, come on, can't you do better than this? Passing 1NT with the South hand is imo unrealistic... dry.gif

LoL. Not without thinking w-a-y too much about this idea. I apologize if I have sounded rude, but the only thing I ever really questioned was the "always" part.

I am well aware of the effectiveness of certain relay systems in the hands of some experts.

[kenrexford]

Cascade, on Aug 21 2009, 08:00 AM, said:

kenrexford, on Aug 22 2009, 12:36 AM, said:

The balanced hand describing his balanced hand serves one function -- what is trumps.

I think it would be possible for the balanced hand to describe more than this. Schemes for showing that information are not commonly played.

But it ought to be possible to show information like: I have a weak doubleton; I have a hand rich in controls etc. Whatever was felt important.

Sure, but even that is responsive.

For example, consider a balanced hand responding to a known two-suiter. The general feel might be "great controls" or "poor controls" depending on the two suits. With KQxx-KQx-Axx-Axx, you love a major two-suiter but hate a minor two-suiter. So, you describe a smile or frown best if you know partner's hand.

True, an A-A-A-A 16-count is better for slam than QJ-QJ-QJ-AQJ before knowing the two suits. Some hands can easily be describes as top-control rich or top-control poor. But, you get far more blended hands, IMO.

That said, a 2NT response to a 1M opening that says "top control poor" or "top control rich" is a workable idea, just like 1M-P-3NT is workable if the bid is one or the other. However, most people just count point ranges and general shape for these calls, and Winston's actual example was a blended hand example, which illustrated my very late point just now arising very well.

[Winstonm]

Ken,

Isn't there another way to do this that involves closely defining the definitions of multiple raises?

For example, in the major suit raise structure I have used you have these parameters: ( Notice that in these gf raise structures no 2/1 or splinter was bid so it is strongly implied that the bid shows a balanced or semi-balanced hand.)

For posterity, this is how it works.

After 1M:

2N: Gf 12-15 and may be 3433 with 4 of the unbid major.

3M=3 or 4-card support, 12-15 with a minimum of two aces but with no more than 4 total control cards (aces and kings combined.)

3C=4-card limit raise or
16+ balanced raise 3 or 4-card support or
Strong jump ***** with 4-card support

3D=3-card limit raise or
gf raise 3 or 4-card support with poor controls (0-2 control cards, but denies as good as AA)

My question is that after such a close definition of the bid why is it still important or necessary for the possibly unbalanced hand to describe his exact shape to the balanced hand?

(It may only be semantics I am talking about - for me, if the auction is in my response structure 1M-2N-3C that is not shape showing but simply an attempt to find the best contract that happens to show partial shape.)

[Winstonm]

helene_t, on Aug 21 2009, 06:24 AM, said:

lol, now I am getting confused, maybe Winston is the one who understood what the issue was and I was wrong. Reminds me of a great post:

Winstonm said:

The purpose of this thread is to insult morons who think they know what the purpose of the thread is. If you want a thread about optimal agreements, start your own.

Anyway, since there are more unbalanced distributions than balanced ones, the balanced hand would need more information to know his partner's exact distribution. But consider this hypothetical bidding system:
- Balanced hands start with a forcing pass, then relay out partner's appr. strength and exact shape.
- Unbalanced hands open and allow partner to relay out their appr. strength and exact shape.

One could compare that to the reverse, and I would predict that the first system would work better.

Perhaps I am being too pragmatic. When I consider "best" I also have to consider things like benefit-to-headache ratio, which means the amount of benefit I perceive in order to put this much energy into remembering a complex and artificial sequence - it best be a LOT.

[kenrexford]

Winston:

First, I love the strong jump ***** bid.

That said, I think you are wildly missing everything there is about bidding slams and about what I have said. If slam bidding were a matter of seeing how many assets you have, combined, then any number of quantitatives would work. Whether that be HCP analysis, or control-count analysis, or LTC analysis, all of this would work. You could even fine-tune this quantitative approach further.

But, slam bidding, IMO, has nothing to do with any sort of quantitative analysis. It best lives in a world of declaring the hand during the bidding. You cannot declare a hand if all you know is how many points or controls or losers dummy has. You need to see the exact cards to know how to play.

Axx-Kxx in dummy is 7 HCP, four losers, three controls, two control cards, one and a half quick tricks. However, if I have QJ10xx-x, I will lose 1.5 tricks on average. If I have x-QJ10xx, I will lose 1 trick and no more.

Axx-Qxx is 6 HCP, 4 1/2 losers, two controls, one control card, one quick trick. If I have KJxxx-x, I will lose 1.5 tricks on average. If I have x-KJxxx, I will lose one trick on average.

What about these situations does the number of controls or shape or anything help?

[Cascade]

kenrexford, on Aug 22 2009, 08:48 AM, said:

However, if I have QJ10xx-x, I will lose 1.5 tricks on average. If I have x-QJ10xx, I will lose 1 trick and no more.

...
If I have KJxxx-x, I will lose 1.5 tricks on average. If I have x-KJxxx, I will lose one trick on average.

I don't understand these bits.

What do you mean?

[kenrexford]

Cascade, on Aug 21 2009, 04:29 PM, said:

kenrexford, on Aug 22 2009, 08:48 AM, said:

However, if I have QJ10xx-x, I will lose 1.5 tricks on average.  If I have x-QJ10xx, I will lose 1 trick and no more.

...
If I have KJxxx-x, I will lose 1.5 tricks on average.  If I have x-KJxxx, I will lose one trick on average.

I don't understand these bits.

What do you mean?

You have agreed that spades are trumps, and you are considering 6♠.

In the red suits, you have Axx in one suit and Kxx in the other suit, either way. If partner has QJ10xx in hearts and a stiff diamond, suppose he gets a heart lead. If you have Axx in hearts but Kxx in diamonds, you have a 50-50 chance of being down one. if you have Kxx in hearts but Axx in diamonds, you probably claim.

The same basic principle is in operation with the other layout.

In other words, the placement of where the cards are is as important as what cards you have.

Here's an easier one, since this concept seems so difficult.

You have 1-5-4-3 shape and have set hearts as trumps. If partner has KQJx in spades and xxx in diamonds, opposite your stiff spade and Axxx in diamonds, life sucks on a diamond lead. If partner instead has xxxx in spades and KQJ in diamonds, life is great. That's why splinters work so well.

So, if partner with 4333 pattern describes his hand as 13-15 with three-card support, and you have a stiff in one suit but Axxx in another, you'd like to be able to describe your stiff so that you and partner can work this all out together, right?

Well, a 3♠ call doesn't give you much room to do that and all the other nice things you might want to do. However, if partner, with say 4333 shape bids 2♣ GF, and you bid 2♦, and partner bids 2♥ setting trumps, now you have more space to work all of this out. Why? Partially because you have kept the auction lower, but also because you have already bid two of your suits and thus have already started painting the picture.

Note, also, that the number of controls outside of diamonds and spades that partner has merely changes what number of controls outside of those suits that you need for this issue to be relevant. So, stating a controls limitation on the 3♠ raise merely determines when you will be screwed by the auction and be able to accurately suspect that fact.

[Winstonm]

Quote

But, slam bidding, IMO, has nothing to do with any sort of quantitative analysis. It best lives in a world of declaring the hand during the bidding.

Ken,

Thanks for answering and you have pretty much said what I suspected was the situation all along - we come from widely divergent schools of thought.

I am of the school that believes bidding should be a dialogue between the two hands, a joint venture to come to an agreement on the best contract.

It seems your view is more like a captaincy method with one hand asking and the other telling. This obviously has merit else the Precision relay system used by what's-their-names years ago wouldn't have been so effective.

Now, to review your analysis:

Quote

You have 1-5-4-3 shape and have set hearts as trumps. If partner has KQJx in spades and xxx in diamonds, opposite your stiff spade and Axxx in diamonds, life sucks on a diamond lead. If partner instead has xxxx in spades and KQJ in diamonds, life is great. That's why splinters work so well.

So, if partner with 4333 pattern describes his hand as 13-15 with three-card support, and you have a stiff in one suit but Axxx in another, you'd like to be able to describe your stiff so that you and partner can work this all out together, right?

Well, a 3♠ call doesn't give you much room to do that and all the other nice things you might want to do. However, if partner, with say 4333 shape bids 2♣ GF, and you bid 2♦, and partner bids 2♥ setting trumps, now you have more space to work all of this out. Why? Partially because you have kept the auction lower, but also because you have already bid two of your suits and thus have already started painting the picture.

Note, also, that the number of controls outside of diamonds and spades that partner has merely changes what number of controls outside of those suits that you need for this issue to be relevant. So, stating a controls limitation on the 3♠ raise merely determines when you will be screwed by the auction and be able to accurately suspect that fact.

You are missing an important concept in the structure I proposed: the 4333 hand is not required to bid 2N or 3M. One of the strengths of the method is flexibility and judgment. xxx, AJ9, KQxx, K109 would bid 2N but Kxx, AJ9, KQxx, xxx could bid 3S or even 2D - nothing is etched in stone nor considered the "perfect bid".

The other point you leave out is what happens when the 5431 hand has a singleton honor? It is all well and good to use xxxx opposite x to show how great splinters work but what about KJxx opposite Q? How well is that judged for 2 winners and not simply 1 loser?

The other issue I have is that slams are comprised of tricks. By bidding 1H-2C-2D-2H yada yada yadi we all get to save room and know we have a heart fit - but if we bid 1H-2C-2D-2H with real clubs opener also knows that responder holds a source of tricks if opener holds say, x, AQJxx, KJxx, Qxx.

The real test of great slam bidding IMO is the ability to sniff out low high card slams based on fit and tricks. Give responder the perfect hand of xxxx, Kxx, A, AKxxx and there is a real chance to reach the 27-point 6H if the club suit is known to be real.

I submit that 1H-2C, 2D-2H, 3C is much more of an effective start to reaching this slam if the club suit is known to be real. It will take a joint effort to visualize the potential for 12 tricks IMO, not one side asking and the other telling.