[Phil]

 semeai, on 2011-July-01, 12:14, said:

Admittedly it's not gaussian, but gwnn's post gets the essential point across. Kudos for the quick code, though; I'm all for actual data over heuristics.

Alternatively, you can just go to wikipedia:

p(16 hcp) = .0331
p(17 hcp) = .0236
p(18 hcp) = .0161

So 16 hcp is more than twice as likely as 18.

Of course, these numbers will change a bit for balanced hands instead of all hands, but presumably not too much.

Added: Impressively, Pavlicek gets it without simulating.

You need to factor in that one person (us) is already know to have a 7 point hand. This is what Hrothgar does above.

[gwnn]

edited out

This post has been edited by gwnn: 2011-July-01, 15:42

[Phil]

ok I'll edit mine too.

[gwnn]

Phil you're really fast! anyways, I'll note in passing that I'm pretty sure hrothgar's original code didn't take it into account and he edited it later, that's why we both just said that it's .33:.16 after hrothgar and the websites too!

[Phil]

 gwnn, on 2011-July-01, 15:47, said:

Phil you're really fast! anyways, I'll note in passing that I'm pretty sure hrothgar's original code didn't take it into account and he edited it later, that's why we both just said that it's .33:.16 after hrothgar and the websites too!

I was referring to his edit.

[whereagles]

 jillybean, on 2011-July-01, 08:49, said:

Playing 16-18nt 2♣ is obvious.

I find it obvious even if 1NT is 15-17

[gwnn]

Not so far

[JLOGIC]

Once upon a time, my partner had 4 spades when he opened 1N, and my hand was a GF and worth far more than 7 points with a singleton and a fit.

Another time, my partner had 18 points, and we made 3N.

An even crazier time, my partner had only a good 17 count, and we were able to make game in NT with just 24 points.

Another time, partner had neither 4 spades nor a good 17-18, and we played 2N with 23-bad 24 HCP and were able to make it, so I lost nothing by inviting.

I told my expert friend about these unusual occurences with my bad 7 count, and he told me that my 7 count was not so bad. In fact, I have 3 9s and a 10 in my 4 card suits. I also have an AJ combination and a Q which are pretty good honor combinations, especially in combination with my good spots.

He even had the crazy idea that everyone who passes with this hand would invite with a similarly average + 4144 8 count opposite a 15-17 NT, which is almost the same thing.

The things you see these days.

[daveharty]

 JLOGIC, on 2011-July-01, 16:42, said:

<sarcasm>

This is funny, and your points are well taken, it's not as if "25 hcp" is some mystical barrier that must be breached in order to make nine tricks in NT. But isn't MP about the frequency of being right? One of my first thoughts on seeing this hand was "3NT is not going to be a favorite here, why not take advantage of our slightly unusual NT range by staying low and hoping to beat the people who start with 1m-1S-2NT...and if partner has 16 or 17, far more likely than 18, other tables will be opening 1NT too and (whereagles comment notwithstanding) won't be getting any higher". Is this twisted?

[semeai]

 Phil, on 2011-July-01, 15:14, said:

You need to factor in that one person (us) is already know to have a 7 point hand. This is what Hrothgar does above.

Yes, definitely good, and the right way to do it. I applaud his post.

In defense of my laziness: As gwnn notes, hrothgar's original post was unconstrained: it used randperm(52), etc at the time. In addition, he had no output at that point, so I just intended to add numbers. At least the ratio p(16hcp) : (18hcp), roughly 2:1, is similar in the constrained and unconstrained data!

[whereagles]

csaba, was that R?

[whereagles]

 JLOGIC, on 2011-July-01, 16:42, said:

The things you see these days.

Still no match for that footage of a circus goat going left and right on a hanging rope, HA!

[gwnn]

whereagles: gnuplot. I will mention in passing that I fitted the points nearly perfectly with two gaussians, one centred around 9.8 and one at 2.3.

[cherdano]

I find this to be a clear 2♣ bid. Most of the arguments have been exchanged, but let me just add that the fear of the natural heart lead against 3NT is way overblown. If we play 3NT, then partner has made a natural 2♥ bid. (I am passing 2♦.)

[the hog]

Clear cut pass for me, 16-18 or not.

[hrothgar]

 semeai, on 2011-July-01, 19:25, said:

Yes, definitely good, and the right way to do it. I applaud his post.

In defense of my laziness: As gwnn notes, hrothgar's original post was unconstrained: it used randperm(52), etc at the time. In addition, he had no output at that point, so I just intended to add numbers. At least the ratio p(16hcp) : (18hcp), roughly 2:1, is similar in the constrained and unconstrained data!

FWIW, as folks have noted, my original post didn't include the conditional probability...
I probably should have waited until I had everything complete to post.

[phil_20686]

I literally cannot imagine passing. When partner has 4 spades I am surely favourite to have a decent game. If he does not we will play in 2/3N and I expect that to be close to neutral equity for that part, with a big gaim when partner has spades,

[hrothgar]

 gwnn, on 2011-July-02, 11:18, said:

I will mention in passing that I fitted the points nearly perfectly with two gaussians, one centred around 9.8 and one at 2.3.

Why would you want to when there are better methods available?

Non parametric fitting techniques are all fine and dandy if you can't specify an appropriate parametric model.
However, its really bad form to use these if you can specify model...

(Of course, the same critique can be leveled at me for using a Monte Carlo simulation instead of solving this all analytically like RP did)

[gwnn]

 hrothgar, on 2011-July-02, 18:31, said:

Why would you want to when there are better methods available?

I was curious how many it would take. Probably a bad excuse.

[Flameous]

Seems like a hand for me to advertise my garbage w/ minor suit stayman twist:

1NT - 2♣ - 2red - 2♠
= Nonforcing minor suit stayman, ie: shows 4 spades and both/long minor.

Not having this toy, it's quite marginal invite. I'd probably still go for it.