[Mbodell]

IMP scoring

---

The MSC write up is in the June magazine, but what would you pick? What do you think does best double dummy? I recognize for many the answer to both might be the same, but it also might not.

There were 7 different cards picked by the MSC panel, but only one got 100 points (picked by half the panel), all others were 50 and below. Without spoiling or biasing too much, the scoring of the problem in MSC and double dummy analysis (at least with the assumptions/restrictions I made) are in conflict.

[rhm]

My copy of the Bridge World has not yet arrived.

I ran a double dummy simulation under IMPs condition:

I assumed:

East has 4 or 5 hearts, 15-17 balanced, a maximum of 3 cards in spades.
West has 4 cards in spades, at most 0-3 cards in hearts. 1-6 cards in a minor and is not 4333.
East-West have a combined HCP between 24 and 30.

Of course you can quibble with these assumptions, but I tell you it is very unlikely to matter significantly.

Under those conditions 3NT is beatable in 1564(31,3% deals out of 5000, but on different opening leads and the opening lead seems to be a pure guess:
In brackets I have computed the success rate of the opening leads.

Low spade does better than ♠T 880 (17,6%) to 767 (15.3%) deals
♥9 or ♥(8) does slightly better than ♥2 (933 (18.7%) to 902 (18%)
Low ♦: 895 (17.9%), higher diamonds do worse 865 (17.3%)♦T and 703 (14.1%) ♦K
♣T is 862 (17.2%), low club is 842 (16.8%) and ♣Q is 816 (16.3%).

Looks to me all not very significant, unless you get fancy and lead the ♦K.
The double dummy simulation hardly confirms the differences in the scores.
Of course when no lead has a success rate of even 20%, overtrick IMPs tend to matter, which in general argues for the more passive leads.

Rainer Herrmann

[nige1]

Mbodell, on 2015-May-28, 02:24, said:

IMP scoring
The MSC write up is in the June magazine, but what would you pick? What do you think does best double dummy? I recognize for many the answer to both might be the same, but it also might not.
There were 7 different cards picked by the MSC panel, but only one got 100 points (picked by half the panel), all others were 50 and below. Without spoiling or biasing too much, the scoring of the problem in MSC and double dummy analysis (at least with the assumptions/restrictions I made) are in conflict.

---

IMO ♥ = 10, ♠ = 9. ♦ = 7. ♣ = 5.
I entered this competition. My initial guess was ♥9, which would have scored a healthy 50%
Unfortunately, I then consulted the Notrump leads section of "Bridge World Standard" which specifies "Spot-card leads: fourth highest; second highest from weak suits", so I corrected my answer to ♥8 -- like PhilKing -- which scored a measly 30%

[PhilG007]

Mbodell, on 2015-May-28, 02:24, said:

IMP scoring

---

The MSC write up is in the June magazine, but what would you pick? What do you think does best double dummy? I recognize for many the answer to both might be the same, but it also might not.

There were 7 different cards picked by the MSC panel, but only one got 100 points (picked by half the panel), all others were 50 and below. Without spoiling or biasing too much, the scoring of the problem in MSC and double dummy analysis (at least with the assumptions/restrictions I made) are in conflict.

Unquestionably,it's got to be 10♠ West seemed disappointed with the 2♥response.He's denied holding five ♠otherwise he would have
asked for a transfer to the suit. Leading ♥ is not to be advised as it would be leading into declarer's suit. And leading from the minor suit honors is more than likely to give away a trick.

[Jinksy]

Without having looked at any of the replies yet (which I'll now do), I would lead a small diamond if partner's X of Stayman would have shown clubs, otherwise I'd lead a small club, on both questions (on the assumption that it would, I've voted for a D atm).

If we're beating this, P's going to need to show up with some values and probably a 5-card suit to set up. On this auction, that's at least as likely to be a minor as a major, and I have much more to offer in the minors (and, correspondingly, he has a less robust suit in expectation to match his lack of overcall over Stayman), so it's a choice of the minors. I imagine he'd X 2♣ on a lot of hands that wouldn't bid 2♦ with the minors reversed, so that's enough for me (subject to above caveat) to risk what would normally feel like a suicidal lead from KTx.

[WesleyC]

I would lean towards a spade lead, although I don't have a strong opinion.

I've spent a lot of time doing double dummy lead analysis, and have noticed a consistent pattern that on marginal hands the simulation prefers a passive lead.

On a few simulations I manually analysed a sample of the hands and concluded that in general, passive leads probably won't fare quite as well in real life. So on an obviously marginal situation like this, I would tend to rely more on the judgement of top human experts than the double dummy sims.

[jodepp]

So much of this type of problem is speculation/guesswork, but here's my 'gut' feeling:

I'm reluctant to lead into declarer's spade suit;
Maybe a top heart through dummy might work;
Either minor could be very right or very wrong.

So, I think at the table I'd try a heart (the 9). Double dummy, I'd guess the ♦10 has the best chance to be the 'killer'.

[WesleyC]

jodepp, on 2015-May-29, 18:24, said:

So much of this type of problem is speculation/guesswork, but here's my 'gut' feeling:

I'm reluctant to lead into declarer's spade suit;
Maybe a top heart through dummy might work;
Either minor could be very right or very wrong.

If your gut knew that declarer had the hearts and dummy had the spades, would it have a different opinion?

FWIW I agree that if declarer's stayman response had been 2S (e.g. dummy showed hearts) then I like a heart lead.

[echo25]

PhilG007, on 2015-May-28, 05:49, said:

Unquestionably,it's got to be 10♠ West seemed disappointed with the 2♥response.He's denied holding five ♠otherwise he would have
asked for a transfer to the suit. Leading ♥ is not to be advised as it would be leading into declarer's suit. And leading from the minor suit honors is more than likely to give away a trick.

No it's not. I would agree if it was MPs, but at IMPs, when opps have points for game, a passive lead isn't good. That's why IMHO a minor lead should be choosen. My choice is diamond 4 or T, depending on if I lead combine or UDCA.

[sfi]

D10 is a standout for me. If it works I score lots of IMPs because they won't find it at the other table, and if not then everyone sympathises and tells me it was a good shot.

Win-win.

[Mbodell]

So in the contest I bid ♠2 with no great certainty. I have slightly different assumptions than Rainer, so my percentages of setting are lower across the board, but mostly the same relative ordering of what is best and what isn't. The MSC 100 point answer was ♦4. My natural guess was that double dummy would actually favor, for instance, a club lead over at the table, because every time your club lead blows a 2 way guess at the table it costs about half a trick, but in double dummy world every 2 way guess the declarer will get right, so it doesn't cost to blow the guess. But I wasn't sure what would be best double dummy in fact, as the passive majors looked appealing, even without the misguess DD boost. Here were the results for 10,000 hands of each card led with number of tricks for declarer, number and percentage of times that we set the contract, and the results IMP'd against "the other table" which is always leading the ♦4 (the number in parens is the per board IMP difference). In terms of number of times we set the contract, it is all pretty close, but in terms of IMP there is about 0.85 IMP/board between best and worst lead double dummy.

Quote

ST lead yeilds 99638 tricks (9.9638), sets 1139 times (11.39%) with imps of -1042 (-0.1042).
S9 lead yeilds 99638 tricks (9.9638), sets 1139 times (11.39%) with imps of -1042 (-0.1042).
S4 lead yeilds 98879 tricks (9.8879), sets 1302 times (13.02%) with imps of 1485 (0.1485).
S2 lead yeilds 98880 tricks (9.888), sets 1302 times (13.02%) with imps of 1484 (0.1484).
H9 lead yeilds 98511 tricks (9.8511), sets 1373 times (13.73%) with imps of 2607 (0.2607).
H8 lead yeilds 98511 tricks (9.8511), sets 1373 times (13.73%) with imps of 2607 (0.2607).
H2 lead yeilds 98707 tricks (9.8707), sets 1319 times (13.19%) with imps of 1817 (0.1817).
DK lead yeilds 102934 tricks (10.2934), sets 1019 times (10.19%) with imps of -5884 (-0.5884).
DT lead yeilds 100395 tricks (10.0395), sets 1213 times (12.13%) with imps of -1251 (-0.1251).
D4 lead yeilds 99821 tricks (9.9821), sets 1270 times (12.7%) with imps of 0 (0.0).
CQ lead yeilds 100925 tricks (10.0925), sets 1055 times (10.55%) with imps of -3233 (-0.3233).
CT lead yeilds 100479 tricks (10.0479), sets 1130 times (11.3%) with imps of -1971 (-0.1971).
C3 lead yeilds 100372 tricks (10.0372), sets 1074 times (10.74%) with imps of -2401 (-0.2401).

The code for the simulation is available at code repository. I did mostly basic restrictions that partner is a passed hand (< 12 no rule of 20 and no preempt), my usual NT bidder restrictions (can have 5M or be 5m422 or 6m322, 15-17 but upgrade some 14 in and a few 17 out) => except NT bidder also needs 4-5 hearts and less than 4 spades due to stayman, dummy has 9-14 hcp, 4 spades, less than 4 hearts, no 6+ minor (with 4x(6+y) I thought 3m would be bid), and partner has no double of stayman (at least loosely defined).

So the ♥9, which received the 50 score on MSC, is best double dummy. Or tied for best with ♥8 (which only got 30 on MSC). Low heart is better than any other suit. Interesting there was one trick difference between the ♠2 and ♠4, which are the best non-heart lead - and MSC gave 40 and 10 respectively, and which are much better than a top spade lead.

[zillahandp]

Dont try to be a hero play with the room, assuming its half competent

[gordontd]

zillahandp, on 2015-May-30, 11:23, said:

Dont try to be a hero play with the room, assuming its half competent

Great aphorism, but what does it mean in this context? Make nine different leads?

[Aardv]

In the double-dummy part of the poll, votes for touching cards should be amalgamated.

On this hand, an active lead can win only if it sets up partner's 5-card or longer suit. That would have to be diamonds, since he didn't double 2♣ (I assume a double would be primarily for the lead). My judgment is that that's just too unlikely.

So we should make the most passive lead. At double-dummy, that's a heart (though it can cost; AJx with partner comes to mind). At single-dummy safety considerations are a bit different - give partner Qxx in both suits and you'd rather lead a heart.

It's interesting that the double-dummy simulations agree that the ten is best from Q10x, but not from K10x. I guess that's because it retains the lead when dummy has Kxx and partner AJxxx.