[RSClyde]

PhilKing, on 2013-September-09, 05:40, said:

Sims are not always what they are cracked up to be. My experience tells me that this is a below average 11-count, and so I simply ignore any synthesised evidence to the contrary!

I think that is completely the wrong way to form a viewpoint.

How many hands "like this" do you pick up? Even hands that feel common, aren't that common.

Do you remember what happens with all of them, or document the results in some way? So you held 300 hands like this last year. Some of the time it makes no difference what you do. So only some of the cases are instructive. Can you give me some idea the number of times it "wasn't really 11"? I mean let's be specific here. If, at the table this hand only really makes 3nt 50% of the time opposite a balance 14, instead of 60%, would you really notice that difference? Just by getting one such hand every couple of days (that's being generous)?


Do you unevenly emphasize some results? (remember disasters, forget mundane hands)


If you begin with a preconception, aren't you incredibly likely to count the hits and explain away the misses? (When it's right to downgrade the response is, "See, that's good bridge, I knew it!" When it wasn't "Well I can't be responsible for that, it was just an unlucky hand".) I mean you admitted yourself that you ignore evidence to the contrary, can't you believe almost anything by that standard?

I mean if a scientist was in charge of testing water quality, would you really want him to say, "Psssh, I've been drinking that water my whole life, it's fine!" The test he was going to run may have had some pitfalls, but anecdotal evidence is hardly a good substitute.

[RSClyde]

cherdano, on 2013-September-08, 18:58, said:

As well as 4333 shape - somehow double dummy defenders find the lead of declarer's doubleton more often.

That would be a valid point except that when I mandated that dummy be 4333 instead of declarer (but maintained 11 opposite 14) 3nt was still at 61%.

[RSClyde]

rhm, on 2013-September-09, 05:26, said:

I agree that K97, QT2, KQJ7, T72 is not a stronger hand, but you might be more interested in getting a diamond lead.
It is fallacy to believe that checking for 3NT is a good test what a hand is worth. That gives too much weight to ace-less flat hands. In suit contracts this is much more of a detriment.
You do not know where you will end up with such hands.

For example I repeated your simulation and used opener's hand opposite a balanced 14 count, but now the opposite hand has 4 spades.
Now you will get to 4♠ and making game drops to 44%, even though I doubt that many would wait for 14 HCP before bidding game when they discover a 4-4 major suit fit. (I used 1000 deals).
The result does not change much, if you manage to get to 3NT if the opposite hand is also 4333.

Rainer Herrmann

If our agreement isn't to open flat 11's then we have a clear pass either way. But if that is our agreement, how can K97, QT2, KQJ7, T72 be an opener but not the other hand? "Sorry partner I know our agreement is to open these hands 1D, but I only had 2 honors in the suit instead of 3, so I took a position."

I don't agree that we'll always get to 4♠ on a balanced hand opposite this on a 4-4 fit. I mean if my partner opened a 14-16 no trump, I'd bid 3nt with that. If I was first to act my partner and I have ways to play 3nt when opener is 4333.

It's a philosophical question, what does it mean for something to be an X count. Goren HCPs seem the best at gauging playability of balanced hands in no trump. So when discussing HCP's, that is what I assume is our barometer.

It's different opening the bidding because you aren't sure what's about to happen. But I hear comments above downgrading hands like this, even in contexts where the commenter, knows that the final contract will be no trump.

[gnasher]

RSClyde, on 2013-September-09, 06:07, said:

How many hands "like this" do you pick up? Even hands that feel common, aren't that common.

From the figures on Wikipedia, and pretending that there's no dependency between shape and high cards, the probability of holding a 4333 11-count is
   (0.6518 - 0.5624) * 0.1054 ~= 0.94%
or about once in every 105 boards.

I think that's sufficient to give each of us enough real data to work with.

Quote

Do you remember what happens with all of them, or document the results in some way?

You may be surprised by Phil's answer.

Quote

I mean if a scientist was in charge of testing water quality, would you really want him to say, "Psssh, I've been drinking that water my whole life, it's fine!" The test he was going to run may have had some pitfalls, but anecdotal evidence is hardly a good substitute.

That's usually true, but it does depend on both the quality of the test and the strength of the anecdotal evidence. Suppose he actually said "100 people have been drinking that water for 100 years, and there's no evidence that any of them has been poisoned by it", and his proposed test was to feed it to 1000 sheep for a week and see what happened. Would you still prefer the test results to the anecdote?

[RSClyde]

gnasher, on 2013-September-09, 08:35, said:

From the figures on Wikipedia, and pretending that there's no dependency between shape and high cards, the probability of holding a 4333 11-count is
   (0.6518 - 0.5624) * 0.1054 ~= 0.94%
or about once in every 105 boards.

I think that's sufficient to give each of us enough real data to work with.


You may be surprised by Phil's answer.


That's usually true, but it does depend on both the quality of the test and the strength of the anecdotal evidence. Suppose he actually said "100 people have been drinking that water for 100 years, and there's no evidence that any of them has been poisoned by it", and his proposed test was to feed it to 1000 sheep for a week and see what happened. Would you still prefer the test results to the anecdote?

Not all 4333 11 counts will involve a meaningful decision. Sometimes your RHO opens 4♠ and it you have an easy call. Sometimes partner has a strong no trump and end up in the same place whether you open or not. We would be talking specifically about hands where deciding to downgrade or not results in reaching (or not) 3nt. As I said earlier I'm not saying that you should open flat 11's, just that you should open this hand if that is your agreement.

But even if you see an example of such a hand every few days, it takes a fair number of trials to estimate proportion, and unless you are writing the results down I'm not sure you're going to sniff out the distinction between .5 and .6.

Deepfinesse, isn't perfect. But it is reasonable to ask why it shouldn't be believed in one case vs another. That is, deepfinesse put 3nt as roughly the same percent contract whether one hand was 4333 or not, why should we be suspicious only of the first case? It may be overly generous to declarer (maybe) but I see no reason to suspect that this only happens when one of declarer's hands is 4333 and no other time.

Everyone has personal experience. I, personally, play lots of bridge hands and never forget anything, I have not observed the phenomenon that 4333 hands are exceptionally weak for playing no trump contracts. So what?

[PhilKing]

RSClyde, on 2013-September-09, 06:07, said:

I mean you admitted yourself that you ignore evidence to the contrary, can't you believe almost anything by that standard?

The short answer is that I said I ignore synthesised evidence to the contrary. Real evidence is another matter altogether.

The long answer will have to wait a while, but you may be surprised at how little you know (believe me when I say I do not mean this offensively).

[VM1973]

RSClyde, on 2013-September-08, 18:04, said:

I'm amazed at how frequently people claim that hands like the opener's "don't evaluate to 11 HCP". This is fairly easy to test.
I Dealmastered a balanced 11 count opposite a balanced 14 count and found that 3nt made 60% of deals.
Now instead of just any 11 count, I used opener's specific hand opposite a balanced 14 count. The result? 62%.
I used 500 hands each way so this is probably within the margin of error for any reasonable alpha value so it probably isn't true that opener holds an above average 11. But it certainly seems to dispel the notion that this hand "isn't really 11".

No matter what, freely rebidding 1nt at reds is beyond terrible. And whether or not you do open balanced 11's is a matter of style/agreement or possibly just generally unwise. But this hand is a real 11 count, whatever that means.

By the way, on the proposed hand of K97, QT2, KQJ7, T72 (which was suggested as possibly being an opener) 3nt opposite a balanced 14, plummeted to 53%.

I think you've missed the point of my post. It's not that the second hand is a better hand for purposes of making 3NT, rather the point was that the original hand has more than one way to be a bad bid than just not having enough power to properly assist a 3NT game opposite a balanced 14 count. Specifically I criticized the bid for A) lack of preemptive power, B) lack of lead-directing power, and C) lack of sacrifice-suggesting power.

Additionally I didn't mention other reservations I have about opening balanced 11 counts willy nilly; namely that against a competent defense that knows that you will invariably open 11 counts, they can more effectively play hands that you pass initially as they can draw the conclusion that you will have, at most, 10 HCPs in a sense of the dog-that-didn't-bark scenario. The same can be said for a 1♦-1[NT]-Pass-3[NT] scenario - they will have no difficulty playing the hand double dummy whereas at other tables where the dealer passed such inferences will not be available.

[RSClyde]

VM1973, on 2013-September-09, 12:03, said:

I think you've missed the point of my post. It's not that the second hand is a better hand for purposes of making 3NT, rather the point was that the original hand has more than one way to be a bad bid than just not having enough power to properly assist a 3NT game opposite a balanced 14 count. Specifically I criticized the bid for A) lack of preemptive power, B) lack of lead-directing power, and C) lack of sacrifice-suggesting power.

Additionally I didn't mention other reservations I have about opening balanced 11 counts willy nilly; namely that against a competent defense that knows that you will invariably open 11 counts, they can more effectively play hands that you pass initially as they can draw the conclusion that you will have, at most, 10 HCPs in a sense of the dog-that-didn't-bark scenario. The same can be said for a 1♦-1[NT]-Pass-3[NT] scenario - they will have no difficulty playing the hand double dummy whereas at other tables where the dealer passed such inferences will not be available.

I don't have a problem with agreeing not to open 11 counts, or agreeing to: either way is fine with me. My issue is with saying that the hand isn't really an 11 count.

I mean I see a philosophical question emerging.

If your agreement is to open 1nt with hands in the following range (a,b) then you open hand x 1nt if and only if it falls within those limits (subject to any upgrading or downgrading you choose to do). Ultimately you view the hand as being worth at least a but not more than b.

Now another day you are playing a different no trump range where you would rebid 1nt with a hand in the range (a,b). Once again you pick up hand x. Is it now automatic that you open 1 of the appropriate suit and rebid 1nt (if appropriate)? As you correctly pointed out the two auctions are different so tactical concerns do apply and hence tailoring to such concerns could cause you to answer "no". So for example, perhaps some would take the view that, if you were playing weak no trumps you'd open 1nt with AKx, Qxxx, Kxx, xxx but playing strong no trumps the thought of opening 1C may be too impalatable and they'd rather pass. It's not my way but I get it.

It's my style to be religious about it. If I don't open 1nt then I don't have a strong no trump. And so similarly if I don't open 1 of a suit then I don't have a weak no trump (whatever our agreed range for that is).

[aguahombre]

PhilKing, on 2013-September-09, 11:13, said:

The long answer will have to wait a while, but you may be surprised at how little you know (believe me when I say I do not mean this offensively).

I always assume that is meant in the nicest way.

[RSClyde]

PhilKing, on 2013-September-09, 11:13, said:

The short answer is that I said I ignore synthesised evidence to the contrary. Real evidence is another matter altogether.

The long answer will have to wait a while, but you may be surprised at how little you know (believe me when I say I do not mean this offensively).

I'm not really getting what about being synthesized is bad. Double dummy analysis could be unrealistic, but why only for 4333 hands?

As far as me personally, I'm not claiming any brilliance here, just telling you what the data says and not seeing a reason to reject it.

[VM1973]

RSClyde, on 2013-September-09, 12:19, said:

I don't have a problem with agreeing not to open 11 counts, or agreeing to: either way is fine with me. My issue is with saying that the hand isn't really an 11 count.

I mean I see a philosophical question emerging.

If your agreement is to open 1nt with hands in the following range (a,b) then you open hand x 1nt if and only if it falls within those limits (subject to any upgrading or downgrading you choose to do). Ultimately you view the hand as being worth at least a but not more than b.

Now another day you are playing a different no trump range where you would rebid 1nt with a hand in the range (a,b). Once again you pick up hand x. Is it now automatic that you open 1 of the appropriate suit and rebid 1nt (if appropriate)? As you correctly pointed out the two auctions are different so tactical concerns do apply and hence tailoring to such concerns could cause you to answer "no". So for example, perhaps some would take the view that, if you were playing weak no trumps you'd open 1nt with AKx, Qxxx, Kxx, xxx but playing strong no trumps the thought of opening 1C may be too impalatable and they'd rather pass. It's not my way but I get it.

It's my style to be religious about it. If I don't open 1nt then I don't have a strong no trump. And so similarly if I don't open 1 of a suit then I don't have a weak no trump (whatever our agreed range for that is).

Well, I used to think exactly as you did. In my case it wasn't opening 11 HCP 4-3-3-3s because I simply don't do that. My concern revolved around opening 1NT with a 5-card major. Obviously there is a chance, when doing so, that the action is wrong but if you open the major you may have rebid problems. My solution was to religiously open 1NT with a 5-card major and 1NT-Pass-3♣ was puppet Stayman to sort it out.

Then I read Kit Woolsey's book Matchpoints. He said it was fine to open 1NT on a hand like:
♠Jxxxx
♥KQx
♦Ax
♣KQx
or
♠AKQxx
♥Ax
♦Kxx
♣xxx
but not with
♠KQxxx
♥Ax
♦Kxx
♣Kxx

His reasoning was that in the first hand the suit in question is too weak to serve as a ready source of tricks at no trump whereas in the second hand the suit is ready to go. In the third hand, however, it will take time to get the suit set up and so it's better to play it in a suit contract.

He explains it far better than I ever could and so I suggest you just pick up a copy of the book and read it.

[cherdano]

So all bridge experts think that an average 4333 hand with 11 hcp is worth less than an average 4432 hand or 5332 hand wiht 11 hcp.
Yet double dummy results show otherwise. Do you think these double dummy results show that all bridge experts are wrong? Then maybe you should base your diet on experiments done with mice. Only.

[RSClyde]

cherdano, on 2013-September-09, 17:12, said:

So all bridge experts think that an average 4333 hand with 11 hcp is worth less than an average 4432 hand or 5332 hand wiht 11 hcp.
Yet double dummy results show otherwise. Do you think these double dummy results show that all bridge experts are wrong? Then maybe you should base your diet on experiments done with mice. Only.

All bridge experts believe that a 4333 11 count is worth less than a 4432 or 5332 hand for playing a no trump contract right? Can you verify this in some way? Not a quote from one or two players, something that indicates that experts are united on this point.

As I've said multiple times now, I'm prepared to adjust my thinking on this if someone simply explains why 4333 hands play better double dummy than other distributions. You suggested an explanation earlier, which was refuted by mandating that dummy, not declarer be 4333 and observing that the success of 3nt remained the same. If this matter is so clear then surely such an explanation exists.

[RSClyde]

VM1973, on 2013-September-09, 16:29, said:

Well, I used to think exactly as you did. In my case it wasn't opening 11 HCP 4-3-3-3s because I simply don't do that. My concern revolved around opening 1NT with a 5-card major. Obviously there is a chance, when doing so, that the action is wrong but if you open the major you may have rebid problems. My solution was to religiously open 1NT with a 5-card major and 1NT-Pass-3♣ was puppet Stayman to sort it out.

Then I read Kit Woolsey's book Matchpoints. He said it was fine to open 1NT on a hand like:
♠Jxxxx
♥KQx
♦Ax
♣KQx
or
♠AKQxx
♥Ax
♦Kxx
♣xxx
but not with
♠KQxxx
♥Ax
♦Kxx
♣Kxx

His reasoning was that in the first hand the suit in question is too weak to serve as a ready source of tricks at no trump whereas in the second hand the suit is ready to go. In the third hand, however, it will take time to get the suit set up and so it's better to play it in a suit contract.

He explains it far better than I ever could and so I suggest you just pick up a copy of the book and read it.

This is probably true playing vanilla 2/1 with a pick-up partner and no inferences. With a rich enough system, I think the value of describing your hand accurately can be more important than masterminding the auction before you have any idea what's happening. Yes, 1nt might not rate to be the best contract if it passes out, but 1♣ doesn't rate to be the right contract if this hand passes out: AKQx, AQxx, Kx, xxx.

[gnasher]

RSClyde, on 2013-September-09, 19:07, said:

As I've said multiple times now, I'm prepared to adjust my thinking on this if someone simply explains why 4333 hands play better double dummy than other distributions. You suggested an explanation earlier, which was refuted by mandating that dummy, not declarer be 4333 and observing that the success of 3nt remained the same. If this matter is so clear then surely such an explanation exists.

Here are some possible explanations:

- If we have a doubleton, the doubleton suit will tend to be the best lead for the defence. If they lead their longest suit, they will tend to find the best lead.
If we have a 4333 shape, the best lead will be more dependent on honour holdings. If they lead their longest suit, it's less likely that this will be their best lead.

- When you're 4333 rather than 4432, you will have more two-way finesse positions. For example, compare A10 opposite KJxx with A10x opposite KJx. Both are worth three tricks double-dummy, but single-dummy the second one has a guess and the first does not.

- When you're 33 in two suits rather than 42, you're more likely to have a guess as to which one to play on.

[RSClyde]

gnasher, on 2013-September-10, 02:18, said:

Here are some possible explanations:

- If we have a doubleton, the doubleton suit will tend to be the best lead for the defence. If they lead their longest suit, they will tend to find the best lead.
If we have a 4333 shape, the best lead will be more dependent on honour holdings. If they lead their longest suit, it's less likely that this will be their best lead.

- When you're 4333 rather than 4432, you will have more two-way finesse positions. For example, compare A10 opposite KJxx with A10x opposite KJx. Both are worth three tricks double-dummy, but single-dummy the second one has a guess and the first does not.

- When you're 33 in two suits rather than 42, you're more likely to have a guess as to which one to play on.

Thank you for the productive dialog.

[rhm]

gnasher, on 2013-September-10, 02:18, said:

Here are some possible explanations:

- If we have a doubleton, the doubleton suit will tend to be the best lead for the defence. If they lead their longest suit, they will tend to find the best lead.
If we have a 4333 shape, the best lead will be more dependent on honour holdings. If they lead their longest suit, it's less likely that this will be their best lead.

- When you're 4333 rather than 4432, you will have more two-way finesse positions. For example, compare A10 opposite KJxx with A10x opposite KJx. Both are worth three tricks double-dummy, but single-dummy the second one has a guess and the first does not.

- When you're 33 in two suits rather than 42, you're more likely to have a guess as to which one to play on.

This is often said, but I am not sure it is very relevant except at the slam level, where declarer typically has more guesses than the defense.
The standard example given is always the two way finesse. But it seems to be finely balanced by options the defense gets right double dummy, for example opening leads.
Why everybody only concentrates on declarer play as an argument that double dummy results are skewed escapes me.
Fact is double dummy derives it tricks differently from single dummy logic, but it does not seem to favor declarer over defenders in general except what is known already, e.g declarer makes on average slightly more tricks in 3NT single dummy than double dummy, because the opening lead tends to be more crucial when there is no trump suit.
What double dummy analysis found out is that generally speaking distribution has on average only a small impact on trick taking capability when there is no trump suit, by which you can control your short suits.
Fit in long suits in notrump are nice but balanced by the fact that you must have short suits then as well.

But distribution does matter a lot and double dummy results agree, provided there is a trump suit for control.

Rainer Herrmann

[gnasher]

rhm, on 2013-September-10, 05:34, said:

Fact is double dummy derives it tricks differently from single dummy logic, but it does not seem to favor declarer over defenders in general except what is known already

We're not talking about whether it favours anyone in general. We're talking about whether it favours anyone more in one specific category of hand than it does in a more general category of hand.

That is, we're comparing

(Single dummy tricks when 4333) - (Double dummy tricks when 4333)

with

(Single dummy tricks when balanced) - (Double dummy tricks when balanced)

This post has been edited by gnasher: 2013-September-10, 05:56

[mcphee]

Pairs and N elects to open, why is not clear to me but so it is. S apparently played a few hands with this partner before and ONLY tried for game with a reasonable opening bid, with a 3 card D fit to boot. To be fair it is not often that I find partner opening the bidding and I need strong NT values in order to have any positive feelings about sniffing for a game.

In pairs the idea is to be conservative, and the south hand was conservative, north was not. Even playing a weak NT the N hand is somewhat suspect holding no aces to open. Taking a second free call of 1NT was serving an active imagination.

N gets it all, from my view.

[JLOGIC]

Assuming they open and freely bid 1N with this kind of hand, then south gets all the blame since he didn't pass 2H which showed a minimum (presumably north has any 11-12 with 3 hearts so what is the point of bidding 3H?). Most likely south didn't get the memo that this is the style, in precision I open the north hand but I don't freely bid 1N. If south thought north had shown something more than a minimum to bid 1N freely, then 3H is certainly fine even after partner shows a minimum in that context...in fact I'd just drive it to game.

So I suspect the problem is with the expectations of the free 1N bid.