[Wackojack]

 32519, on 2012-March-30, 00:42, said:

Hi Frances

Partner and I were playing a very basic version of Puppet Stayman which concentrated on the majors (little effort was made to include minor suit orientated hands from responder).

This thread was started when we started asking “why all this PS fuss?” The times when either of us held a 5332 20-21 HCP hand was seldom. More often than not we ended up playing in a 4-4 major suit fit, or a 5-3 major suit fit with responder holding the 5-card suit not the opener. We got there via Jacoby transfer bids.

So I looked up the frequency of occurrence of typical 20-21 HCP NT hands. This is what I found: The probability of being dealt –
1. 5332 and 20-21 HCP = 0.08% (5-card major, minor 5-card suits excluded)
2. 4432 and 20-21 HCP = 0.08%
3. 4333 and 20-21 HCP = 0.12%
...Total.........................= 0.28%

The probability of being dealt a bust –
1. 4450 and 0-4 HCP = 0.01% Club shortness (losing Garbage Stayman)
2. 4441 and 0-4 HCP = 0.07% Club shortness (losing Garbage Stayman)
3. 4405 and 0-4 HCP = 0.01% Diamond shortness (losing Crawling Stayman)
4. 4414 and 0-4 HCP = 0.07% Diamond shortness (losing Crawling Stayman)
5. 4423 and 0-4 HCP = 0.17% Diamond shortness (losing Crawling Stayman)
...Total.....................= 0.33%
Less benefit of 5332..= (0.08%)
Net benefit/loss.........= 0.25%

Having no mechanism to improve the auction when responder has a bust, we were in a net loss situation.

Seeing the numbers, partner and I decided to dump PS. The majority of the face-to-face bridge we play is match points. The decision to dump PS became even easier when we were getting a top score in 3NT making 10 tricks when the rest of the field were making 10 tricks in a major.

I respect your decision for playing PS. I’m just saying why I dumped it.

I have some doubts about the efficacy of puppet Stayman. However, like many posters, I find the way you argue the case against is particularly unconvincing. The post on frequencies above is a case in point.

I will leave aside the probabilities you gave on getting 0-4 HCP opposite a balanced 20-21 which I think are wrong. More importantly, they are meaningless. If you want to do figures to compare frequency of opportunity to do garbage or crawling stayman against opening 2NT with a 5 card major 5332, then you should compare proportions.

So the proportion of hands 5 card major 5332's compared with all other balanced hands is about 7.7%.

Now looking at the possible Garbage and Crawling Stayman hands you have given, we have to assume that they will be the same relative frequency opposite a 2NT opening bid as they would be with with unknown other hands. This should be roughly true. Then proportion of:

5440's with club or diamond shortages = 0.24%
4441's with club or diamond shortage = 1.5%
4432's with majors = 3.6%

Thus 5.3% of all distributions are suitable for G or S Stayman.

So with your logic (which of course I don't accept anyway) the benefit of using Puppet is 7.7%- 5.5% = plus 2.2%

[manudude03]

Does anyone really play garbage/crawling stayman over a 2NT opener? Surely the biggest loss of puppet stayman is losing the smolen-type hands, of which some variations can handle anyway.

[campboy]

For anyone who might be interested, there are 3393199716 balanced 20-21 counts, of which 534618234, so about 15.8%, have a 5-card major.

[johnu]

 32519, on 2012-March-30, 00:42, said:

...................................
...Total.....................= 0.33%
Less benefit of 5332..= (0.08%)
Net benefit/loss.........= 0.25%

I don't understand your numbers. Besides that, you seem to be assuming that every time you could use crawling Stayman you are improving your expected value. Overall, 2NT is a far better contract than 3 of a major on a 4-3 fit when you don't find a 4-4 fit. Finding a 4-4 major fit when you are 4-4 in the majors is supposed to be about 52%, so you'll end up in 4-3 fit (or less) about 48% of the time.

My simulations on Dealmaster Pro show that the gains you have from finding a good 4-4 fit are approximately offset by the losses you have when you end up in a 4-3 fit. And if opener has a 2=2-4-5 or 2=2-3-6 semi balanced hand with minors, you won't even have a 4-3 fit to play. You also have a loss when you can't use Smolen.

[1eyedjack]

 campboy, on 2012-March-31, 10:01, said:

For anyone who might be interested, there are 3393199716 balanced 20-21 counts, of which 534618234, so about 15.8%, have a 5-card major.

I am interested. Of those 15.8%, what proportion find a responder with precisely 3 card support, and fewer than 4 in the other major?

[32519]

 Wackojack, on 2012-March-31, 06:47, said:

I have some doubts about the efficacy of puppet Stayman. However, like many posters, I find the way you argue the case against is particularly unconvincing. The post on frequencies above is a case in point.

I will leave aside the probabilities you gave on getting 0-4 HCP opposite a balanced 20-21 which I think are wrong. More importantly, they are meaningless. If you want to do figures to compare frequency of opportunity to do garbage or crawling stayman against opening 2NT with a 5 card major 5332, then you should compare proportions.

So the proportion of hands 5 card major 5332's compared with all other balanced hands is about 7.7%.

Now looking at the possible Garbage and Crawling Stayman hands you have given, we have to assume that they will be the same relative frequency opposite a 2NT opening bid as they would be with with unknown other hands. This should be roughly true. Then proportion of:

5440's with club or diamond shortages = 0.24%
4441's with club or diamond shortage = 1.5%
4432's with majors = 3.6%

Thus 5.3% of all distributions are suitable for G or S Stayman.

So with your logic (which of course I don't accept anyway) the benefit of using Puppet is 7.7%- 5.5% = plus 2.2%

These numbers of yours here are very close to my own numbers for a 0-37 HCP range. Obviously we cannot use this range. The 2NT bid requirement is 20-21 HCP. I used 0-4 HCP as the bust hand requirement.

The numbers you have quoted fall away hopelessly once you start specifying the HCP range.

[32519]

 johnu, on 2012-March-31, 11:57, said:

I don't understand your numbers. Besides that, you seem to be assuming that every time you could use crawling Stayman you are improving your expected value. Overall, 2NT is a far better contract than 3 of a major on a 4-3 fit when you don't find a 4-4 fit. Finding a 4-4 major fit when you are 4-4 in the majors is supposed to be about 52%, so you'll end up in 4-3 fit (or less) about 48% of the time.

My simulations on Dealmaster Pro show that the gains you have from finding a good 4-4 fit are approximately offset by the losses you have when you end up in a 4-3 fit. And if opener has a 2=2-4-5 or 2=2-3-6 semi balanced hand with minors, you won't even have a 4-3 fit to play. You also have a loss when you can't use Smolen.

Puppet Stayman was created to allow a 5-card major within your 2NT range. Why would you want to allow 2245 or 2236 minor suit orientated hands in your 2NT range and still play Puppet Stayman? For sure I see no need for Crawling Stayman anymore when responder holds a bust.

With a 2245 or 2236 minor suit orientated hand opposite a bust, you now face the risk of the opponents running both of your 2-card suits?

[32519]

A lot of numbers are starting to form part of this thread. Allow me to add some more why I believe Puppet Stayman is unnecessary.

22.56% = The probability of responder being dealt 5M(any)(any)(any) and 5-20 HCP (a positive response)
32.02% = The probability of responder being dealt 5+M(any)(any)(any) and 5-20 HCP (a positive response)
00.08% = The probability of opener being dealt 5M332 and 20-21 HCP.

Just looking at the numbers, anyone other than top flight players, gain little through adding to the memory load on a 0.08% probability that opener’s 2NT contains a 5-card major. Without transferring into responders 5-card major suit immediately you start adding additional bids into your Puppet sequence to “right side” the contract at a higher level (that is if your Puppet sequence contains such bids).

I will be extremely surprised if non top flight players continue with PS.

[1eyedjack]

 32519, on 2012-April-01, 02:35, said:

00.08% = The probability of opener being dealt 5M332 and 20-21 HCP.

Something looks a bit off here. Surely the question to pose is what is the pobability of opener being dealt 5M332 GIVEN that he has 20-21 (or whatever) points AND is known to be any 4333 or 4432 or 5332 (or perhaps 6m322 or 5422)
That has to be considerably more than 0.08%. I am more inclined to believe Campoy's 15%. And then my question remains unanswered: the chance of responder having precisely 3 card support will reduce that somewhat to a meaningful figure for the purpose of deciding whether it is worth devoting resources.

[campboy]

 1eyedjack, on 2012-March-31, 13:11, said:

I am interested. Of those 15.8%, what proportion find a responder with precisely 3 card support, and fewer than 4 in the other major?

I assume you meant "4 or fewer" rather than "fewer than 4" since that is what seems to matter.

My program doesn't do two hands at once, but so long as you don't care about responder's pointcount it's easy to calculate. Once you fix opener's hand with 5 spades (say), the number of hands for responder with 3 spades and h hearts is
⁸C₃*^13-kC_h*^18+kC_10-h (out of ³⁹C₁₃ total hands)
where k is opener's heart length. When k=3 there is a 26.1% chance of 3 spades and <5 hearts; when k=2 it is a 23.8% chance. k=3 is twice as likely as k=2 so the overall probability is 25.3%.

Putting those two figures together, you miss a 5-3 major fit on less than 4% of 2NT openers by not having puppet available. On some of the hands we've included here responder will not bid puppet anyway, either because he is too weak or because he judges 3NT will be the best spot even if there is a 5-3 fit. It is also possible that he will decide to bid something else on some minor-oriented hands.

[1eyedjack]

you are right i meant 4 or fewer. thanks

[32519]

 32519, on 2012-March-30, 00:42, said:

I looked up the frequency of occurrence of typical 20-21 HCP NT hands. This is what I found: The probability of being dealt –
1. 5332 and 20-21 HCP = 0.08% (5-card major, minor 5-card suits excluded)
2. 4432 and 20-21 HCP = 0.08%
3. 4333 and 20-21 HCP = 0.12%
......Total.............................= 0.28%

 1eyedjack, on 2012-April-01, 03:19, said:

Something looks a bit off here. Surely the question to pose is what is the pobability of opener being dealt 5M332 GIVEN that he has 20-21 (or whatever) points AND is known to be any 4333 or 4432 or 5332 (or perhaps 6m322 or 5422)That has to be considerably more than 0.08%. I am more inclined to believe Campoy's 15%. And then my question remains unanswered: the chance of responder having precisely 3 card support will reduce that somewhat to a meaningful figure for the purpose of deciding whether it is worth devoting resources.

I don’t have DealMaster Pro. I am using BBO’s deal generator. Setting up the hand specifications in BBO’s deal generator: Minimum / Maximum as follows –

5S / 5S
2H / 3H
2D / 3D
2C / 3C
20 HCP / 21 HCP

The “Odds” gives me a 0.04% probability of being dealt such a hand. So I need to double this figure for the other major, which gives me my 0.08%. I can improve this % slightly as follows –
5S / 13S
2H / 13H
2D / 13D
2C / 13C

The “Odds” now improve to 0.08%. So I need to double this figure for the other major, which now gives me a 0.16% probability of being dealt a hand containing a 5-card major and 20-21 HCP. This slightly improved % is skewed by the fact that the 5-card major may in fact now be a 7-card major.

[campboy]

 32519, on 2012-March-30, 00:42, said:

1. 5332 and 20-21 HCP = 0.08% (5-card major, minor 5-card suits excluded)
2. 4432 and 20-21 HCP = 0.08%
3. 4333 and 20-21 HCP = 0.12%
...Total.........................= 0.28%

#1 is right. #2 is wrong; it should be 0.24%. #3 is right. The total is wrong both because #2 is and also because you didn't include 5332 hands with a 5-card minor and those are still automatic 2NT openers.

[blackshoe]

 32519, on 2012-April-01, 02:35, said:

I will be extremely surprised if non top flight players continue with PS.

Then you will be extremely surprised. <shrug>

[Wackojack]

 32519, on 2012-April-01, 02:35, said:

Just looking at the numbers, anyone other than top flight players, gain little through adding to the memory load on a 0.08% probability that opener’s 2NT contains a 5-card major.

I will be extremely surprised if non top flight players continue with PS.

Simply not true. If you look up the suit probability distributions, you will see that the probability of getting a 5332 distribution is 15.5%. Therefore the probability that this will be a major suit 5332 is half that i.e. 7.75%.
Now look at all balanced distributions:
P 4333 = 10.5%
P 4432 = 21.5%
P 5332 = 15.5%
Total = 46.5%

Thus the extra chance of getting a major 5332 distribution given that you have opened 2NT = 7.75/46.5 = 16.7%

So if you believe that you need to find out if partner has a 5 card major, then it occurs for 16.7% of all 2NT opening bids. Very well worth it.

[awm]

You're not really using the right measure for memory here. For example, I'm sure the use most people put 2NT-3♠ to (typically some kind of minor suit slam try) are much less frequent than puppet stayman. Yet I don't find any even semi-serious partnerships that leave that 3♠ bid wholly undefined.

Puppet is only a little more complicated than regular stayman (which has a lower frequency-of-use actually, since you usually need a 4M to use it), and it's in a place where most people play methods. It's not like Lebensohl or transfer advances of overcalls (for example) which "look like natural bids" and could be easy to forget (and thus might not be worth playing except that they are very frequent and also very useful).

I just don't think "memory load" is a big concern with regard to puppet stayman. Personally I don't play it when it's my choice (will play it if partner insists), but that's because I don't think it's a good method, not because it's hard to remember.

[johnu]

 32519, on 2012-April-01, 01:58, said:

Puppet Stayman was created to allow a 5-card major within your 2NT range. Why would you want to allow 2245 or 2236 minor suit orientated hands in your 2NT range and still play Puppet Stayman? For sure I see no need for Crawling Stayman anymore when responder holds a bust.

With a 2245 or 2236 minor suit orientated hand opposite a bust, you now face the risk of the opponents running both of your 2-card suits?

What would you bid open with something like:

---

I don't know anybody in my bridge circles who would open anything except 2NT if it shows 20-21. The fact that you open a semi-balanced hand like this with 2NT is unrelated to whether you decide to play puppet Stayman. It doesn't seem like a good idea to base your bidding system on partner having a bust

[phil_20686]

 johnu, on 2012-April-01, 16:22, said:

I don't know anybody in my bridge circles who would open anything except 2NT.

Really? Not know many strong club players?

But seriously, this thread seems to have argued about ridiculous things. Who cares about frequency? It is your expected gain that matters, and the fact is that the big gain from puppet stayman is when responder has a 3 card major and a singleton and you find a 53 fit when its a better spot than 3N. That is not uncommon.

Running out to a better partscore by playing garbage stayman is a joke surely? I mean I have basically given up non-forcing auctions after 2N. You either pass it or you go to game, because every sequence after a 2N opener is valuable, and having NF bids cuts down the number of ways I can describe my hand.

The big loss of puppet stayman is that it becomes much more difficult to make slam tries in your 4-4 major fit. After 2N-3c-3d-3H showing 4S, partner bids 3N so you discover that your fit is actually hearts, and there are no bids left for making a slam try. If you shoehorn 4m in as a slam try then you are giving up bidding with 4M5m or 4M6m hands in a natural way. You also have to find something else to do with the 54 major hands. These are not massive losses but they do cost.

I mean, in general slam hands are less frequent than game hands, but you still need system for them. If you limited yourself to never bidding slams PS is obviously a win, as you basically never prefer to play in 5m rather than 3N when partner is 20-22 balanced, so I could use those bids for my extra major hands.

The idea that I might want garbage stayman is just LOL.

[32519]

 Wackojack, on 2012-April-01, 11:28, said:

Simply not true. If you look up the suit probability distributions, you will see that the probability of getting a 5332 distribution is 15.5%. Therefore the probability that this will be a major suit 5332 is half that i.e. 7.75%.
Now look at all balanced distributions:
P 4333 = 10.5%
P 4432 = 21.5%
P 5332 = 15.5%
Total = 46.5%

Thus the extra chance of getting a major 5332 distribution given that you have opened 2NT = 7.75/46.5 = 16.7%

So if you believe that you need to find out if partner has a 5 card major, then it occurs for 16.7% of all 2NT opening bids. Very well worth it.

Simply not true. These figures are for a 0-37 HCP range which you obviously cannot use. Go to BBOs deal generator and fill in the hand constraints and calculate the “Odds” of being dealt each hand type.

My Last Puppet Post
Let us close shop on the Puppet Stayman discussion here. Enough has been said. The pro Puppet camp will continue playing it. The anti Puppet camp isn’t likely to start playing it now. The undecided have enough information to make their decision into which camp they want to fall.

[campboy]

 Wackojack, on 2012-April-01, 11:28, said:

Now look at all balanced distributions:
P 4333 = 10.5%
P 4432 = 21.5%
P 5332 = 15.5%
Total = 46.5%

Thus the extra chance of getting a major 5332 distribution given that you have opened 2NT = 7.75/46.5 = 16.7%

So if you believe that you need to find out if partner has a 5 card major, then it occurs for 16.7% of all 2NT opening bids. Very well worth it.

The reason that this figure is wrong is that pointcount is not independent of shape. Hands with very high (or low) pointcount are likely to be flatter.

(There is also the issue that 10.5+21.5+15.5=47.5 as well as the fact that to guarantee 1dp accuracy in your final answer you need rather more accuracy in your initial figures. You should get 16.3% for 0-37 as opposed to 15.7% for 20-21.)