Simple judgement call (corrected)

[the hog]

Jeff Goldsmith (who wrote the tool natch), has remarked in his puzzle serious that the adjustments KnR makes to a hand are not net netural. Using KnR to evaluate hands and then comparing to HCP is not a good plan.

 

What is a "tool natch" and a "puzzle serious"?

[Cthulhu D]

What is a "tool natch" and a "puzzle serious"?

 

Whoops, I meant 'puzzle series' and tool naturally!

 

:)

[nige1]

Probably most people who insta pass here don't play classical 15-17 1NT but popular 14+-17 one which is considerably weakier on average thus makes passing more attractive option.

This is very good 8. It makes 3NT opposite 16-17 hcp balanced without 5M 53% of the time dd which translates to probably about 58-60% in practice

Please would bluecalm or somebody else help me. A few years ago, I wrote a couple of dealer scripts but I don't know how to include a double-dummy analysis. I don't have GIB but I believe there is some way of using another DD solver. My initial crude attempt below fails

 

/****** ******* ******* ******* ******* ******* ******* *******
   	Dealer script: 8never.ds: 10 May 2012: Nigel Guthrie.
   	North has a specific 8-count.
   	South has a 15-17 notrump opener.
   	Is discretion the better part of valour?
   	For Dealer by Hans van Staveren & Henk Uijterwaal
   	[url="http://www.bridgebase.com/tools/dealer/dealer.php"]http://www.bridgebas...aler/dealer.php[/url]
******* ******* ******* ******* ******* ******* ******* ******/
produce 100
predeal  north SAJ8, HJ965, DQ974, C82
condition shape (south, any 4333 + any 4432 + any 5332) and 
   hcp (south) > 14 and hcp (south) < 18
action frequency "Tricks: " (tricks (south, notrump), 0, 13)
# End of Dealer script 8never.

[P_Marlowe]

Stayman + invite.

[jdeegan]

:P El Paso, no problemo. True you might have a combined 25 HCP, BUT the frequency of 15, 16 and 17 HCP hands for partner isn't 1/3, 1/3, 1/3. The probability of the 17 HCP hand is LESS than 1/3, and you generally need the 17 HCP hand just to have a reasonable play for 3NT, and you are not even vul. You have imo an average eight count with no obvious source of tricks.

 

By the way, let me pile on a bit in my disdain for KnR. I am sure it means something in some bidding context, but I don't know what that might be. Everyone knows that the Milton Work count is defective in that it undervalues aces, overvalues jacks and doesn't properly account for the power of high card combinations. There are maybe four or five legitimate ways to evaluate a bridge hand in the context of an auction. I don't see how KnR fits in, plus I don't carry a computer to the card table.

[bluecalm]

In general I try to avoid inviting. The invite gives information to the opposition and gets us to silly contracts like 2NT and 3♥.

 

I concede that it might be the way to go. I am still not convinced somehow, mainly because if there is an edge in this approach it's much smaller than what you get if you just land in the same contract than other table(s). With perfect players though I think the chances are that you are right. I wait for the day when computers could play at strong human level so we can just feed million hands like that to them and see if pass or 2C is better here.

[Statto]

By the way, let me pile on a bit in my disdain for KnR. I am sure it means something in some bidding context, but I don't know what that might be. Everyone knows that the Milton Work count is defective in that it undervalues aces, overvalues jacks and doesn't properly account for the power of high card combinations.

KnR works better for unbalanced hands where you end up playing in a suit contract, scoring on a par with Zar points. If both hands are fairly balanced and you play in NT, Milton Work is actually more effective than KnR, and probably doesn't undervalue Aces much if at all, though does undervalue well placed 10s and 9s.

[nige1]

I still can't simulate the actual hand using DEALER because I can't get the tricks function to work; so I downloaded and ran

Richard Pavlicek's fantastic deal-finder program and database of over two million deals with double-dummy analysis for all denominations and declarers

to "simulate" a flat 8 HCP with an ace, a queen and two knaves opposite a flat 15-17 HCP.

 

123 SP>14 & SP<18 & SF<3 & NP=8 & NA=1 & NQ=1 & NJ=2 & NF<3 & SM>8 # Rotate 2M+ deals & count 3NT makes 

Disappointingly for me, this "simulation" overwhelmingly confirms the majority viewpoint :( :( :(

• 15 HCP: 1446 deals of which 299 make 3N
• 16 HCP: 1161 deals of which 408 make 3N
• 17 HCP: 826 deals of which 440 make 3N
• 15-17 HCP: 3433 deals of which 1147 make 3N

[Statto]

Disappointingly for me, this "simulation" overwhelmingly confirmed the majority viewpoint :( :( :(

Did you specify 2 nines and 2 eights (but no tens) in the simulation? It may make a difference. Thanks for the link :)

[nige1]

Did you specify 2 nines and 2 eights (but no tens) in the simulation? It may make a difference. Thanks for the link :)

66 deals satisfy the criteria, South can make 3N in 24 of those cases :( :( :(

 

 123 SP>14 & SP<18 & SF<3 & NP=8 & NA=1 & NQ=1 & NJ=2 & NT=0 & N9=2 & N8 =2 &NF < 3 & SM>8 # Pip restrictions.

Each restriction on the North hand drastically reduces the number of deals selected by DEAL-FINDER form its pre-analysed database.

DEALER suffers less from this problem because it takes the designated North cards out of the deck, before pseudo-randomly dealing the remaining cards.

[bluecalm]

Please would bluecalm or somebody else help me. A few years ago, I wrote a couple of dealer scripts but I don't know how to include a double-dummy analysis. I don't have GIB but I believe there is some way of using another DD solver. My initial crude attempt below fails

 

I am using dmpro for all my simuls. I have no idea how dealer scripts even work :)

[MrAce]

In general I try to avoid inviting. The invite gives information to the opposition and gets us to silly contracts like 2NT and 3♥. Thus I try to narrow my invite range as much as possible, either bidding game or passing.... so basically my invite is a "good 8 to a bad 9" and I would pass with most 8s and force game with most 9s. This doesn't quality as a good 8 for me; in fact I think it is a slightly below-average 8. Pass.

 

I love this way of thinking.

 

This also protects you from some nasty doubles in a game contract after an invitation auction, which they would never do if you blast game since you can be much stronger than what you have. As well as some rare but possible partscore doubles.

[bluecalm]

Damn, i am really curious about this non-invite style.

My intuition really doesn't like it but every time I tried to run some simuls it seemed that invite is always the worst option.

Is there any semi-decent card playing bridge program which could be use for simulating this ? Or any open-source semi-decent bridge program ? I could do bidding myself, just need automatic card player.

 

Let's say we have:

Ax T9x AJxx xxxx do you think it's just a bash opposite 15-17 1NT ? My intuition strongly tells me I will be collecting a lot of 0's if I start playing this way.

[Statto]

Did you specify 2 nines and 2 eights (but no tens) in the simulation? It may make a difference. Thanks for the link :)

66 deals satisfy the criteria, South can make 3N in 24 of those cases :( :( :(

That sort of suggest to me that this hand is in the invite zone, and pretty close to the punting zone vul at teams if you don't have an invite. B-)