[blml] Detroit NABC+ Case 1 -- What do you bid and why?

[Adam Wildavsky]

Thanks to everyone who replied! For those who haven't yet looked the
complete writeup is here:

  http://web2.acbl.org/casebooks/Detroit2008/01-NABC+.pdf

This case gave me fits. It's the most difficult I've seen in a while.
My original question may have sounded like a poll, but my intent,
which I hope I accomplished, was to ask "What, if anything, does the
UI suggest and why?"

Even after reading all your comments, and those from NAC members and
private correspondence, I could not answer the question to my own
satisfaction. On that account I broke down and did something I've been
putting off for years. I bought a copy of Dealmaster Pro and learned
how to use it. This software has a simulation module combined with the
Deep Finesse double-dummy solver. It allows one to generate random
deals that fit a set of criteria and then learn the theoretical
optimal result when played at various contracts. I entered the actual
North hand and criteria for each of the other three hands in two
cases. The first was when South holds 19 HCP or fewer, roughly
matching the AI case, and then second when South has 20 HCP or more,
roughly matching the UI case. For the UI case I also added that South
would not be 4333, since with that shape he would never have a problem
over 1NT.

My goal was to learn whether the UI made balancing more or less
attractive. This would be directly relevant only if players at the
table were also computers, but in a sense this kind of calculation is
something every player does, to one degree or another. In any case I
found the analysis useful.

Here are the criteria I used in both the AI and UI cases:

East holds 15-17 HCP, balanced, with no 5 card major, no 6 card minor,
and not 5422. East also does not hold both four hearts and 17 support
points, since with that he'd likely have pre-accepted.

South is balanced. Further, if South hold 16 HCP or more he holds at
least 3 hearts, since otherwise he'd have doubled 2 hearts.
South does not hold five spades, since then he might have bid over 2H.

West hold five or six hearts. West is not 4540 since he'd have likely
used Stayman. West does not hold both 4 spades and 8+ distribution
points, since he might have used Stayman. West does not hold 6 hearts
with two of the top three honors -- he might have invited. West does
not hold 6 hearts with 9+ distribution points -- again a possible
invite.

Distribution points are 4321 HCP plus 321 for void/stiff/doubleton.

I know that in the AI case I didn't need to restrict South to 19 HCP
or fewer, but it makes little practical difference since the 20+ HCP
hands are rare.

I evaluated the results of 2H versus 3C for two cases, one where 3C
ends the auction and another where 3C is always doubled when it's
going down. Fortunately none of our opponents are that good, but this
lets us generate figures for opponents who get the decision right a
percentage of the time.

I'll post the results after giving you all a chance to review my
criteria and suggest changes.

AW