[Mbodell]

---

Imps, 3♠ shows 4 or 5 spades and fewer than 4 hearts. You play BWS defensive agreements. Choose the touching cards if you lead any one of the touching, but in the 2nd best dd, don't choose the same answer you made for 1st best. 8 of the 13 cards were led by BWS Panel members.

[nige1]

Mbodell asks "Imps, 3♠ shows 4 or 5 spades and fewer than 4 hearts. You play BWS defensive agreements. Choose the touching cards if you lead any one of the touching, but in the 2nd best dd, don't choose the same answer you made for 1st best. 8 of the 13 cards were led by BWS Panel members."

I rank
1. ♠T. Passive and unimaginative.
2. ♦8. Seemingly passive but might help declarer to pick up ♦s.
3. ♣T. Might surrender the 9th trick.
4. ♥7. Likely to help declarer pick up ♥s

---

[WesleyC]

Having read the Bird/Anthias books and also done a bunch of double dummy simulations myself, I would lead a diamond here with a fair degree of confidence.

1st point is things looks extremely grim for our side. We hold only 4 HCP, partner's HCP are sitting under declarer and hearts are splitting evenly. Based on similar situations I'd estimate our chance of defeating the contract at around 10%.
2nd point is that we will almost never be able to defeat 3NT out of our own hand. With RHO holding 4/5 spades and our hand having only 1 outside entry, the Spade suit is an illusion. In order to beat 3NT we're going to require (probably 4) tricks in partner's hand.
3rd point is that partner had an opportunity to double 3C for a lead and didn't.

Once you can accept these main points, a diamond lead stands out as by far the best chance of hitting partner's 5c suit and beating the contract.

[phil_20686]

WesleyC, on 2015-November-02, 19:25, said:

Having read the Bird/Anthias books and also done a bunch of double dummy simulations myself, I would lead a diamond here with a fair degree of confidence.

1st point is things looks extremely grim for our side. We hold only 4 HCP, partner's HCP are sitting under declarer and hearts are splitting evenly. Based on similar situations I'd estimate our chance of defeating the contract at around 10%.
2nd point is that we will almost never be able to defeat 3NT out of our own hand. With RHO holding 4/5 spades and our hand having only 1 outside entry, the Spade suit is an illusion. In order to beat 3NT we're going to require (probably 4) tricks in partner's hand.
3rd point is that partner had an opportunity to double 3C for a lead and didn't.

Once you can accept these main points, a diamond lead stands out as by far the best chance of hitting partner's 5c suit and beating the contract.

While I broadly agree I feel that:

1) Partner is only marginally less likely to hold 5h or 5C.

2) When he does hold five decent cards then the presence of an honour massively increases your odds of setting up the suit.

3) However, your side will also need tempo's if teh suit isnt cashing, and the club holding is of the kind which might block a suit, so I would choose a heart over a club.

I mean, if you hid partner with AQxxx hearts your chances of setting this are much higher than if you hid AQxxx diamonds. I think that weights the decision back towards a heart. Although I voted for a diamond I think that its pretty close between a diamond and a heart, especially with the info that rho has 4 spades and not 4 hearts.

[WesleyC]

I assumed from they stayman bid that LHO will have a 4c heart suit.

[Jinksy]

Club for me. P needs very little to bring the suit in (far less than he'd need to X 3♣), and I have no reason to think the opps don't have a surfeit of points, so a passive lead looks like giving up. I expect it to do even better DD, where P will be able to not block with Qxxxx and xxx in dummy.

[Mbodell]

phil_20686, on 2015-November-03, 04:46, said:

While I broadly agree I feel that:

1) Partner is only marginally less likely to hold 5h or 5C.

WesleyC, on 2015-November-03, 09:57, said:

I assumed from they stayman bid that LHO will have a 4c heart suit.

I agree with Wesley, 5h is quite a bit less likely (nearly 0) because of the 4 card suit on our left and the 2NT bidder on our right having at least 2 and we have 3.

5c is possible, but I think with 5 good clubs a X might have happened, although maybe not with just AQ222 which is very likely good enough opposite our KT9.

[Thiros]

I made ♣10 my choice. We have a balanced hand with 4 HCP and so hitting partner's strength looks like our best hope (it isn't like we will set up and run the spades). The ♣10 lead will be money if partner can turn up with ♣QJxxx and a side entry, which is possible even if dummy has extras. We could hardly expect partner to double 3♣ with that (what if opener had ♣AK10x?). Yes the club lead will often hand declarer the ♣Q, but some of the time the gift will only be an overtrick (I would lead a low ♠ at matchpoints).

The ♦8 lead is indeed the likeliest to find partner with a five-card suit but in order to produce a set it would have to be very strong. More frequently it will simply give declarer a beacon while slicing through partner's broken holding. I know that if it were me at the table and I led a diamond with this hand, dummy would put down ♦1092 and declarer would have ♦AQ64.

A heart lead has the same issues. It does hit the occasional ♥KQ9x with a side ace, but with dummy known to have four hearts, more likely to give declarer a pickup that he wouldn't make on his own.

[WesleyC]

Thiros, on 2015-November-04, 23:43, said:

The ♦8 lead is indeed the likeliest to find partner with a five-card suit but in order to produce a set it would have to be very strong. More frequently it will simply give declarer a beacon while slicing through partner's broken holding. I know that if it were me at the table and I led a diamond with this hand, dummy would put down ♦1092 and declarer would have ♦AQ64.

I think the flaw in this idea is that you're assuming if the diamonds are ♦T9x, ♦AQ64 we might have missed a chance to beat 3NT. If you run into this worst case diamond layout, your lead didn't matter. Nothing mattered, declarer was always going to roll home with some large number of tricks.

You hold 4 HCP, the opponents free bid to 3NT, all of partner's HCP are sitting under declarer, hearts are breaking evenly - you KNOW from the start that you're rarely going to win.

On a hand like this where you need to start with the assumption that you are a huge underdog without much to lose. Take risks and think about the positive cases rather than the negatives. You've got nothing to lose!

[Mbodell]

For some results, here was how the MSC voted:

Quote

Action  Score  Panel Votes  % of Solvers
S4       100      10               8
ST        90       4              26
S2        90       2               0
S9        90       1               0
D8        80       6              40
CT        70       3              13
HJ        60       1               2
H6        60       1               7
CK        20       0               0
other      0       0               4

There are only 11 votes from bbf at the time of this voting, but 6 are for the diamond, 2 each for top spade and club T/9, and 1 for the 2 of spade, none for the top scoring 4th best spade.

For best lead DD bbf forum have 8 votes for diamonds, 2 for clubs, and 1 for low spade. For 2nd best DD lead it is very split with 3 for low heart and 2 each for top spade, 4th best spade, diamond, and club T/9.

When I simmed this, the results were extremely close. Close enough that I ran the results 100,000 times rather than the usual 10,000 times just to cut down on the sample error (the other errors - like differing assumptions about the hands, and the difference between SD and DD - no doubt dwarf the sample error, but wanted to make sure. As always simulation code available for inspection). The results were:

Quote

ST lead yeilds 1032794 tricks (10.32794), sets 25637 times (25.637%) with imps of 0 (0.0).
S9 lead yeilds 1032794 tricks (10.32794), sets 25637 times (25.637%) with imps of 0 (0.0).
S8 lead yeilds 1032794 tricks (10.32794), sets 25637 times (25.637%) with imps of 0 (0.0).
S4 lead yeilds 1038494 tricks (10.38494), sets 24006 times (24.006%) with imps of -14759 (-0.14759).
S2 lead yeilds 1038486 tricks (10.38486), sets 24007 times (24.007%) with imps of -14705 (-0.14705).
HJ lead yeilds 1045175 tricks (10.45175), sets 23034 times (23.034%) with imps of -33345 (-0.33345).
H7 lead yeilds 1038501 tricks (10.38501), sets 24905 times (24.905%) with imps of -16846 (-0.16846).
H6 lead yeilds 1038501 tricks (10.38501), sets 24905 times (24.905%) with imps of -16846 (-0.16846).
D8 lead yeilds 1031908 tricks (10.31908), sets 25582 times (25.582%) with imps of 8659 (0.08659).
D7 lead yeilds 1031908 tricks (10.31908), sets 25582 times (25.582%) with imps of 8659 (0.08659).
CK lead yeilds 1086238 tricks (10.86238), sets 16297 times (16.297%) with imps of -82069 (-0.82069).
CT lead yeilds 1041469 tricks (10.41469), sets 23680 times (23.68%) with imps of -12816 (-0.12816).
C9 lead yeilds 1041469 tricks (10.41469), sets 23680 times (23.68%) with imps of -12816 (-0.12816).

I originally thought based on a small sample run that top spade would be best, and indeed it leads to the set slighlty more often than diamond (about 1 in 2000 hands more), but when you look at the imps, the diamond lead ends up leading to the fewest tricks and hence is actually the lead that averages the best IMP score. However it is extremely close as usually the difference in leads is measured in .25 or .3 imp between best and second best. Here it is less than .1 IMP. And in fact the spread from best to worst lead here, leaving out the major "hero" lead and minor "hero" lead of ♣K and ♥J is just around .25 IMP. Even the disaster lead like the club K is less than a 1 IMP loss, usually the disaster leads are more like 1.5 IMP losses. You might think it is because we set it so rarely, but actually 20-25% is pretty reasonably frequent. And in the July problem we set the hand more like 5-10% but the imp differences were still quite large. So really I think it is because the leads are so similar in their results with only 2% really between any lead outside the top honors (and even the ♥J is only an extra 0.5% make and about .15 IMP worse than low heart).

Interesting that the general readers lean so strongly for the double dummy best lead, while the panel plurality favors the beginners rule of "4th from longest" and splits nearly 2:1 in favor of some spade.

[WesleyC]

Thanks for posting your simulation code - it's a lot more readable than mine!

I'd change a few things (although there is no guarantee they would effect the final result).

1) I think you should include the possibility for the 2NT opener to have a stiff Q, K or A especially in a 4441 shape. My style would be to include only a minor singleton.

2) I'm possibly mistaken, but are you assuming that north will double stayman on any hand with 6+ clubs? Are you also suggesting that they wouldn't double 3C holding AQJxx of clubs without 10 total HCP? Seems wrong.

3) I like the idea of your 'upgrade_nt' proc, but I would add a provision to downgrade for tight honours, a less strong emphasis on just aces/tens/nine and bigger focus on the texture in the long suit (rather than 10's 9's in the whole hand).

4) Restricting West to 5+ HCP is too conservative. West has no choice but to investigate game on most shapely hands with 3-4 HCPs - eg [x KTxx xxx 9xxxx] or similar because the upside of finding a heart fit is too great.

[Thiros]

WesleyC, on 2015-November-07, 07:23, said:

I think the flaw in this idea is that you're assuming if the diamonds are ♦T9x, ♦AQ64 we might have missed a chance to beat 3NT. If you run into this worst case diamond layout, your lead didn't matter. Nothing mattered, declarer was always going to roll home with some large number of tricks.

You hold 4 HCP, the opponents free bid to 3NT, all of partner's HCP are sitting under declarer, hearts are breaking evenly - you KNOW from the start that you're rarely going to win.

On a hand like this where you need to start with the assumption that you are a huge underdog without much to lose. Take risks and think about the positive cases rather than the negatives. You've got nothing to lose!

I'm not so sure that's the issue here. I think the issue is our choice of what to hope for.

Whatever we lead has to hit substantial length and strength with partner if we are to get a plus score on this hand (alternatively we can lead a spade, and take the position that either it will turn out to be a slow hand, or only overtrick IMPs will be at stake). A diamond lead would have to hit, I would say at the very least, ♦KQ10xx (or ♦KJ109x or ♦KJ10xxx, with queen and one in dummy) plus a side entry (without which declarer could freeze partner's suit out by ducking once). A club lead would have to hit ♣QJxxx plus a side entry (on a few rare layouts ♣Qxxxx may be enough). A heart lead would have to hit at least ♥KQ9x plus a side ace and necessitate that declarer has to force out our ♣K for his nine tricks. I lead club because it requires of partner the fewest HCP.

[Mbodell]

WesleyC, on 2015-November-09, 02:39, said:

2) I'm possibly mistaken, but are you assuming that north will double stayman on any hand with 6+ clubs? Are you also suggesting that they wouldn't double 3C holding AQJxx of clubs without 10 total HCP? Seems wrong.

I re-ran a version with no restrictions on partner (I.e., pretend they would never double even with AQJxxxx of clubs). The results were what you'd expect, mostly the same or slightly worse for everything except a little better for club leads now. So now dimaonds are best, then club T/9, then top spade. The results were still super tight.

Quote

ST lead yeilds 1035958 tricks (10.35958), sets 25214 times (25.214%) with imps of 0 (0.0).
S9 lead yeilds 1035958 tricks (10.35958), sets 25214 times (25.214%) with imps of 0 (0.0).
S8 lead yeilds 1035958 tricks (10.35958), sets 25214 times (25.214%) with imps of 0 (0.0).
S4 lead yeilds 1041869 tricks (10.41869), sets 23604 times (23.604%) with imps of -15373 (-0.15373).
S2 lead yeilds 1041865 tricks (10.41865), sets 23604 times (23.604%) with imps of -15325 (-0.15325).
HJ lead yeilds 1047859 tricks (10.47859), sets 22640 times (22.64%) with imps of -29984 (-0.29984).
H7 lead yeilds 1041302 tricks (10.41302), sets 24390 times (24.39%) with imps of -14423 (-0.14423).
H6 lead yeilds 1041302 tricks (10.41302), sets 24390 times (24.39%) with imps of -14423 (-0.14423).
D8 lead yeilds 1035204 tricks (10.35204), sets 25111 times (25.111%) with imps of 9331 (0.09331).
D7 lead yeilds 1035204 tricks (10.35204), sets 25111 times (25.111%) with imps of 9331 (0.09331).
CK lead yeilds 1081564 tricks (10.81564), sets 17161 times (17.161%) with imps of -63626 (-0.63626).
CT lead yeilds 1038673 tricks (10.38673), sets 24118 times (24.118000000000002%) with imps of 2676 (0.02676).
C9 lead yeilds 1038673 tricks (10.38673), sets 24118 times (24.118000000000002%) with imps of 2676 (0.02676).