[gnasher]

jdonn, on Dec 27 2009, 08:21 AM, said:

I don't think that's a fair comparison. Justin's point is that on such a hand you could add any high cards and you would have slam. But opposite your example you could have a good slam opposite useful high cards that constitute a minimum strength opener, it would just be bad opposite (the admittedly more likely case of) wasted strength.

True, except that there are some ways to add high cards to Justin's example which make it not a 2♦ bid. Would opener bid 2♦ with x xx AKJxxx KQxx?

[gnasher]

mike777, on Dec 27 2009, 09:37 AM, said:

ok I dont get any of these posts

thought 4c as said long ago...was easy.

Are you saying that because Justin said he'd bid 4♣ the rest of us should just agree and move onto the next problem?

[mike777]

gnasher, on Dec 27 2009, 05:08 AM, said:

mike777, on Dec 27 2009, 09:37 AM, said:

ok I dont get any of these posts

thought 4c as said long ago...was easy.

Are you saying that because Justin said he'd bid 4♣ the rest of us should just agree and move onto the next problem?

God in heaven...


NO


I thought 4c was easy ...even for me.....a ..nonexpert...

[gnasher]

eyhung, on Dec 27 2009, 03:06 AM, said:

My point about the unfairness of the 8-count example is that given that partner has a 1♦ opener, it requires a somewhat pathological distribution of high cards and shape in the side suits to make 5♦ poor.

Do you mean that my examples of real opening bids with ♠Qxx or worse tend towards the worst case (in the same way that a 1246 shape with ♦AK tends towards the best case)? That's what they were intended to do: my point is that we may well be able to find out how close partner's hand is to the worst case, before deciding whether to play at the five-level.

Quote

Yes, a spade singleton is unlikely -- but so is xxx given that partner has an opener and the opponents have failed to overcall or double.  I expect a spade control or spade values very frequently, and the sim bears this out.

It's certainly true that at this stage of the auction ♠Hxx or ♠HHx is more likely than ♠xxx, both because there are more ways to have a spade honour than not, and because partner opened the bidding. If, however, partner has such a holding, and bids notrumps twice, we won't often belong in 6♦. Or are you going to say that Kxx x AJxxxx KQx is also "pathological"?

Quote

And even if 4C completely forecloses notrump, you're only missing out a small percentage of the time.

Without accepting the premise that your simulation is a reasonable representation of reality, I would say that even if we belong in 3NT on only 7.3% of hands (67 / 913), that possibility is worth exploring. It's not as though we have to make a final decision now: there are nearly two whole levels of bidding between 2♦ and 3NT.

Quote

I suspect trying 2♠ with the intent of stopping in 3NT on some auctions is going to lead to a missed diamond slam more frequently.

Rather than dealing in suspicions and generalisations, why don't you provide some examples of hands where opener would bid 2NT followed by 3NT, and we would miss a good slam? As I understand it, you have dealt yourself 690 hands where we can make a slam. You could have a look at some of them. If the risk of playing in 3NT with 6♦ on is so great, it shouldn't take you long to find a few convincing examples.

Quote

When a small slam is making approximately two-thirds of the time, it seems unwise to attempt to cater to 3NT especially when it's unclear to me that partner is making an intelligent decision on 3NT vs. diamonds.  Part of this is because you have not told me what your 2♠ - 3♦ - 3NT sequence shows -- but I doubt most people really know either.  How does partner know that you don't hold something like AKxx Jxxx Q987 A, where heart weakness is death; or AJ AKxx Q9xx xxx where club weakness is death?

The sequence I suggested was 2♠-2NT;3♦-3NT.

Whilst the two hands you give are possible, neither is typical. When I bid hearts, spades and diamonds, partner should assume that I have length in hearts, spades and diamonds, and therefore shortage in clubs. When I support diamonds in a game-forcing sequence, I suggest that I am suitable for a high-level diamond contract.

If partner bids 3NT over 3♦, he should have a double club stop. If he has weaker clubs, and therefore values in one of the majors, he should bid where his values are. Isn't that just bridge?

Quote

Partner is just rebidding 3NT with 0-2 hearts and stuff in the blacks.  If your blacks fit well, sometime you're playing in 3NT when you're cold for 5 or 6D, and sometimes you're even down in 3NT when 5D is cold.  I don't think finding a good 3NT is nearly as frequent as finding a good 6D. 

If you can come up with a plausible sequence that could describe your hand so well that a "random expert North" could accurately judge 3NT vs. 6♦, then your approach has merit,

Over 3♦, partner doesn't have to make a choice between 3NT and 6♦. He has to decide between "3NT immediately", and "not 3NT immediately". If he does the latter, we will then continue our conversation about where to play the hand.

Quote

but absent that, I think your hand needs to take control.

This is a side issue, but does 4♣ "take control"? A splinter usually means "This is my shortage; please evaluate your hand opposite that shortage." If you bid 4♣, partner will think that what you're looking for is well-fitting major-suit cards. If I wanted to take control, I'd bid 4♦, which basically forces partner to bid 4♠ any time that he has spade control.

Quote

I reran the sim

I have some comments on this too, but I'll put them in a separate post.

[mike777]

Again
1) 4c over 2d seems easy
2) No No NO bidding a splinter very often means not being captain on the hand.
3 I am going to assume after 4c .......anyone bidding 4h is rkc for D.

[eyhung]

gnasher, on Dec 27 2009, 06:05 AM, said:

That's what they were intended to do: my point is that we may well be able to find out how close partner's hand is to the worst case, before deciding whether to play at the five-level.

And my point is that we can't.

You've now described what you think 1♦ - 1♥ - 2♦ - 2♠ - 3♦ - 3NT should show : double club stop. While a few hands with a double club stop (Axx Q AJxxxx KJx) provide good play for slam, I agree that if you are on the same wavelength where partner will only bid 3NT with lots of club wastage, you are likely to make a good decision on 3NT vs 6♦. But, it doesn't seem to me that a random expert partner would restrict his 3NTs to just that: what if he holds something like Kx Qx AKxxxx KTx or Ax Qx AJTxxx QJT. No double stop, but he has to bid something over 3♦, and I would be surprised if 3NT didn't get some votes in an expert panel. Opposite these experts, you'd pass 3NT and 6♦ is great in either case. And your suggested interpretation doesn't get you to stop in 3NT opposite some of the hands you listed such as Qxx x AKJxxx Qxx (rebid 3♠) or even Kxx x AKJxxx Kxx (rebid 3♠). Face it, you have a primed-out gf hand with a 10-card diamond fit, it's going be really hard to stop in 3NT at IMPs without foreclosing some 6♦ contracts unless you've had some discussion on machinery to do this.

Quote

even if we belong in 3NT on only 7.3% of hands (67 / 913), that possibility is worth exploring.  It's not as though we have to make a final decision now: there are nearly two whole levels of bidding between 2♦ and 3NT.

My point is that up until your last post, it was completely unclear that your use of those two levels is going to be fielded by partner appropriately. I did say it might be possible to use the space between 2♦ and 3NT to design sequences where you could make an intelligent decision between 3NT and diamonds, but the OP was looking for a practical answer.

Quote

This is a side issue, but does 4♣ "take control"?

Re: taking control, I meant : forcing the contract past 3NT and saying "we're playing diamonds dammit, how does your hand fit opposite club shortness?", that is, controlling the strain and pushing the partnership to 5♦. Apologies for the confusion -- clearly a splinter relinquishes control of the bidding to partner -- but it does control the strain, and that's what I was talking about.

[y66]

Good problem. I'd have bid 4C at the table. But I really like the idea that ... 2S 2N 3D 3N also tells us that opener has wasted values in clubs and that, since he didn't bid 3S over 3D, spades could be a problem.

[gnasher]

eyhung, on Dec 27 2009, 08:39 PM, said:

You've now described what you think 1♦ - 1♥ - 2♦ - 2♠ - 3♦ - 3NT should show : double club stop.

That's not merely what I think it should show: it's what I think it does show.

Quote

While a few hands with a double club stop (Axx Q AJxxxx KJx) provide good play for slam,

If I had Axx Q AJxxxx xxx and had heard partner show a game-force with hearts, spades, and diamond support, I'd be thinking that 5♦ was likely to be cold, and 6♦ a distinct possibility. Adding ♣KJ doesn't make the hand any less suitable for diamonds.

Furthermore, KJx opposite a singleton isn't a double stop. It would be ridiculous to go down in 3NT with Axx Q AJxxxx KJx opposite KQx AJxxx Qxxx x.

All in all, it seems right for opener to go past 3NT himself with this.

Quote

But, it doesn't seem to me that a random expert partner would restrict his 3NTs to just that: what if he holds something like Kx Qx AKxxxx KTx or Ax Qx AJTxxx QJT.  No double stop, but he has to bid something over 3♦, and I would be surprised if 3NT didn't get some votes in an expert panel.

Both of those hands look like 1NT openers to me, though that wouldn't help us to get to 6♦.

If I had fortuitously opened 1♦ on either of these, I'd bid 3♥ over 3♦. Qx in partner's first suit seems like something he might want to hear about, and I've already declined to raise and then declined to give preference over 2♠, so he won't expect any more than Hx. I think that 3NT with either of these hands would be a pretty poor bid, regardless of how many of your hypothetical experts would vote for it.

Quote

And your suggested interpretation doesn't get you to stop in 3NT opposite some of the hands you listed such as Qxx x AKJxxx Qxx (rebid 3♠) or even Kxx x AKJxxx Kxx (rebid 3♠).

True - bidding is an imperfect science, and it's hard to distinguish between a singleton ace and a small singleton. But if I get to 3NT opposite some of the hands where it's right, and still don't miss any good slams, that appears to be an improvement on never getting to 3NT when it's right.

Quote

Face it, you have a primed-out gf hand with a 10-card diamond fit, it's going be really hard to stop in 3NT at IMPs without foreclosing some 6♦ contracts unless you've had some discussion on machinery to do this. 

Quote

My point is that up until your last post, it was completely unclear that your use of those two levels is going to be fielded by partner appropriately.  I did say it might be possible to use the space between 2♦ and 3NT to design sequences where you could make an intelligent decision between 3NT and diamonds, but the OP was looking for a practical answer.

I'm not talking about machinery, or anything else that requires design or prior discussion. All I'm suggesting is a sequence where the players exchange information using bids that mean what they sound like they mean, and jointly arrive at a decision about the right contract. What on earth is impractical about that?

[gnasher]

eyhung, on Dec 27 2009, 03:06 AM, said:

I reran the sim with tighter constraints :

These are the constraints after tightening? I think you were wise not to publish your original criteria.

Quote

North has 6+ diamonds, 12-bad 16 HCP, no side 4-card suits

This appears to mean that opener will
- Pass or open 2♦ on all 10- and 11-counts that contain 6+ diamonds.
- Never open 1NT on a 6322 14-16 count.
- Never temporise with 2♣ on a goodish 1363.
- Never raise to 2♥ with x36x.
- Never rebid 2♣ with 6-4 in the minors.

I realise that styles vary somewhat from place to place, but do you really know people who bid like this?

Quote

East cannot act over 1D
West cannot act over 1H

Can you tell me what the constraints for these are?

I hope you don't think I'm being unreasonably picky, but when you provide an apparently scientific analysis, you should expect it be treated like science. If you want people to place any weight on your simulations, you should provide not just the results but also the method used to obtain them, so that your audience can review, criticise and attempt to replicate it.

[eyhung]

Quote

I hope you don't think I'm being unreasonably picky, but when you provide an apparently scientific analysis, you should expect it be treated like science.  If you want people to place any weight on your simulations, you should provide not just the results but also the method used to obtain them, so that your audience can review, criticise and attempt to replicate it.

I didn't realize this was a scientific review journal. My goal in simulation is to provide a ballpark figure, not a significant number down to the last decimal point that caters to all sorts of bidding styles. Modeling hands which would bid 2♥ on x36x, 2♣ on xx63, and the like feels like more work than the result is worth, as everyone has their own personal heuristics for these auctions. I was trying to use a set that nobody would argue with: you certainly wouldn't be surprised if partner showed up with a x36x hand that you would have raised to 2♥ with. And if a reasonable set of constraints tells me that 6♦ is on the order of 2/3 likely to make, and that 3NT is not likely to be the right contract over 5♦, then unless you can construct auctions that can convincingly determine that slam in diamonds is wrong below 3NT, I don't see the point in fooling around with 2♠ when game in diamonds is practically certain (90%+), which argues for forcing to 5♦ and giving yourself the best practical chance to investigate 6♦. Playing opposite an expert partner you trust, you can try 2♠, but my feeling is that getting into a subtle auction where you can count on partner to rebid 3♥ with Kx Qx in the majors opposite a hand that could hold ♠Txxx just leads to too many missed diamond slams.

FWIW, my 1♦ openers include 11-counts, I should have stated that. Basically any hand that satisfies rule of 20 and 2 quick tricks. 10-counts, too, but that's impossible if opener can't have a 4-card side suit.

As for the "quick-and-dirty" first run sim, the earlier results were based on not putting any constraints on the enemy hands, which I felt actually increases the chance of a spade singleton and thus a favorable opener for slam. I guess this was mitigated by specifying that West was a passed hand and East was a passed hand over 1♦, so opener rates to have more strength.

[eyhung]

gnasher, on Dec 27 2009, 03:50 PM, said:

Quote

East cannot act over 1D
West cannot act over 1H

Can you tell me what the constraints for these are?

### non-overcall over 1D
### doesn't handle two-suiters yet
proc pass_1D {hand} {
 [space] [space]if {[preempt $hand]} { return 0 }
 [space] [space]if {[spades $hand] >= 5 && [hcp $hand] >= 8} { return 0 }
 [space] [space]if {[spades $hand] >= 6 && [hcp $hand] >= 5 && [ok_lsuit $hand]} { return 0 
}
 [space] [space]if {[hearts $hand] >= 5 && [hcp $hand] >= 8} { return 0 }
 [space] [space]if {[hearts $hand] >= 6 && [hcp $hand] >= 5 && [ok_lsuit $hand]} { return 0 
}
 [space] [space]if {[clubs $hand] >= 5 && [hcp $hand] >= 12 && [good_lsuit $hand]} { return 
0 }
 [space] [space]if {[hcp $hand] >= 15} { return 0 }
 [space] [space]if {[diam_dbl $hand 13]} { return 0 }
 [space] [space]return 1
}

### non-overcall over 1H
### doesn't handle two-suiters yet
proc pass_1H {hand} {
 [space] [space]if {[preempt $hand]} { return 0 }
 [space] [space]if {[spades $hand] >= 5 && [hcp $hand] >= 8} { return 0 }
 [space] [space]if {[spades $hand] >= 6 && [hcp $hand] >= 5 && [ok_lsuit $hand]} { return 0 
}
 [space] [space]if {[clubs $hand] >= 5 && [hcp $hand] >= 12 && [good_lsuit $hand]} { return 
0 }
 [space] [space]if {[diamonds $hand] >= 5 && [hcp $hand] >= 12 && [good_lsuit $hand]} { retu
rn 0 }
 [space] [space]if {[hcp $hand] >= 15} { return 0 }
 [space] [space]if {[heart_dbl $hand 13]} { return 0 }
 [space] [space]return 1
}

[diam_dbl] is essentially a hand that has 13 support points (1/3/5 for shortness) counting diamond shortness, and valuing unsupported soft diamond honors as 0. So Axxx Kxxx x Kxxx would double 1♦, but Axxx Jxxx Q Kxxx would not. [heart_dbl] is similar but evaluates heart shortness. The support point number is an argument to handle takeout doubles at higher levels.

[ok_lsuit] and [good_lsuit] use Kleinman's berry suit evaluator, OK is 15 berries or better, Good suits are 18 berries or better. http://dannykleinman...37%20POINTS.pdf for more info on berries. I should collapse all the suit evaluators to use berries as an argument, but I just discovered the berries idea recently.

One of my pet projects is to come up with a reasonable set of heuristics to model early actions to use in simulations and then contrast with heuristics supplied by other leading players. I haven't had as much time as I like to do this. You are clearly an excellent player, Andy, so if you want to share your heuristics with me on when you should open 1NT with a 6cm, for example, we can take this to e-mail and I can build a simulator that will "bid like you".

[Old York]

4♦ seems like the best bid (under the circumstances), as it was your partner's responsibility to either open or rebid 1NT, so no blame can be laid at your doorstep. To treat 4♦ as non-forcing seems absurd

Rebidding 4♣ could be a total disaster with an unknown partner, and a splinter bid should deny a singleton honour (esp Ace), so it makes it impossible for opener to evaluate his hand

The "practical" rebid is 3NT, because opponents seem to have a 9 card club fit, and will probably lead them. Even if they find a spade lead, the suit may be blocked
I don't like trying to find minor suit slams with unknown partners who are likely to pass 4♣ or 4♦.

Tony
p.s. I would open 1♣ as West, or double after 1♦-1♥. West's passes made the hand tricky