Lead problem H - MSC July

[Mbodell]

IMP, all white

 

[hv=pc=n&s=sk9862h873d98cj74&d=e&v=0&b=14&a=1np2cp2sp3nppp]133|200[/hv]

 

6 different cards were chosen by panelists.

[benlessard]

I think the best lead and the best DD lead is easy. Its the 2nd best DD lead that is tough.

[Free]

I'm torn between ♦ and ♣. ♦ has more chance of finding partner's 5 card suit and partner didn't double 2♣, but still I prefer 3 card leads over doubletons, especially if that 3 card suit contains an honour. Finding 4♣s in partners hand can be enough, finding only 4♦s in partners hand may be a disaster.

 

DD, dunno, whatever ;)

[benlessard]

♦ has more chance of finding partner's 5 card suit

this and the fact that both opps showed a 4M wich increase the odds the D break 3-3.

[PhilKing]

It's difficult to guess the "double dummy" answers without knowing the bidding assumptions made by the person running the simulation.

[Mbodell]

It's difficult to guess the "double dummy" answers without knowing the bidding assumptions made by the person running the simulation.

 

Usually any reasonable assumptions will lead to similar ordering of the results, even if different assumptions will lead to different percentages of makes or what not. As in, the assumptions usually tilt the hand more to how likely over all leads the hand is to make, and not rock too much the overall order of the hand. At least for reasonable assumptions.

 

I've done the assumptions in the code available here. Most of the assumptions are "bog standard" for this auction although I did add that East can't have 4 hearts for this sequence, since this is from BWS bidding and with both majors one is supposed to bid 2♥ over stayman for BWS. Also, partner can't have 6+ clubs (or else double stayman) and also can't have both 10+ hcp and 5+ clubs (or else also double stayman) - this certainly isn't 100% accurate (probably be worth capturing suit quality in clubs better), but seemed a reasonable swag to me - but certainly might have hurt clubs vis-a-vis these results (I invite anyone else to dd with their own assumptions). The results I got were:

 

SK lead yeilds 108225 tricks (10.8225), sets 103 times (1.03%) with imps of -16452 (-1.6452).

S9 lead yeilds 102306 tricks (10.2306), sets 467 times (4.67%) with imps of -7789 (-0.7789).

S8 lead yeilds 102306 tricks (10.2306), sets 467 times (4.67%) with imps of -7789 (-0.7789).

S6 lead yeilds 102268 tricks (10.2268), sets 508 times (5.08%) with imps of -7405 (-0.7405).

S2 lead yeilds 102266 tricks (10.2266), sets 510 times (5.1%) with imps of -7384 (-0.7384).

H8 lead yeilds 99766 tricks (9.9766), sets 753 times (7.53%) with imps of -2760 (-0.276).

H7 lead yeilds 99766 tricks (9.9766), sets 753 times (7.53%) with imps of -2760 (-0.276).

H3 lead yeilds 99866 tricks (9.9866), sets 739 times (7.39%) with imps of -2984 (-0.2984).

D9 lead yeilds 99261 tricks (9.9261), sets 1011 times (10.11%) with imps of 0 (0.0).

D8 lead yeilds 99261 tricks (9.9261), sets 1011 times (10.11%) with imps of 0 (0.0).

CJ lead yeilds 101494 tricks (10.1494), sets 590 times (5.8999999999999995%) with imps of -5881 (-0.5881).

C7 lead yeilds 100492 tricks (10.0492), sets 719 times (7.19%) with imps of -3788 (-0.3788).

C4 lead yeilds 100493 tricks (10.0493), sets 722 times (7.22%) with imps of -3760 (-0.376).

 

This makes the diamond 9/8 the top lead and the heart 8/7 the second best lead.

 

At the master's solvers clubs result page (currently at here - although it will roll over soon):

 

Action     Score     Panel Votes     % of Solvers
D9          100         14               53
S2           90          3                2
S6           80          6                9
S8           80          1                1
S9           80          0                1
H8           60          3               20
C4           50          2                9
else          0          0                5

 

So the panel and solvers have about half finding the double dummy "right" lead. The clear second most popular solvers lead is also the second best double dummy lead, but it scores not as well as the spades lead (even if you combine all the spades). It is interesting that the scoring is quite a bit more generous to this problem than to the previous month (covered in a thread here) where the 2nd best score was 50 out of 100 for the lead that was best double dummy. If anything, the difference in IMP between the different choices this month was slightly larger than the past month, but yet the non-100 scoring was much more generous than the previous month. It is also interesting that as of this post no one in this thread had argued for any spade lead either as what they'd do at the table or as the guess for first or second best double dummy lead.

[benlessard]

I expected D to be best and S to be worst (IRL too) however I had no clue if clubs was going to be better than heart.

 

Anyone got an idea why H do better than clubs ? Is it the same if the suit is 843 ?

[gnasher]

Anyone got an idea why H do better than clubs ? Is it the same if the suit is 843 ?

 

 

A club lead will sometimes cost a trick by force, whereas a heart can at worst save declarer a guess. The effect of that is exaggerated by using double-dummy analysis, but I'd expect the heart lead to work better single-dummy too.

 

[benlessard]

So this weight more than the fact that partner is more likely to have lenght in clubs than lenght in hearts ?

[gnasher]

So this weight more than the fact that partner is more likely to have lenght in clubs than lenght in hearts ?

 

Yes, because the chance of finding partner with five clubs is much less than the chance of finding him with some holding where the lead costs a trick.

 

There are two ways to beat 3NT:

- Set up enough tricks to beat it. That usually means leading a five-card suit, or your partner's five-card suit.

- Lead passively and wait for declarer to go down.

 

Leading from Jxx is about as far away from passive as you can get, but it's also not very likely to hit a five-card suit in partner's hand. The number of distributions where partner has five clubs is much less than number where he doesn't.

Edited June 30, 2015 by gnasher

[nige1]

[hv=pc=n&s=sk9862h873d98cj74&d=e&v=0&b=14&a=1np2cp2sp3nppp]133|200|IMP, all white6 different cards were chosen by panelists.[/hv]

Against notrump contracts... Spot-card leads ... second highest from weak suits.

After rechecking the official leading agreements, I led ♥7, which was worth zero points :( This was the second consecutive MSC quiz, where the "top-of-nothing" false-card scored more than the systemic "second-highest" lead :(

Infamy! Infamy! They've all got it in for me!

SK lead yeilds 108225 tricks (10.8225), sets 103 times (1.03%) with imps of -16452 (-1.6452).

S9 lead yeilds 102306 tricks (10.2306), sets 467 times (4.67%) with imps of -7789 (-0.7789).

S8 lead yeilds 102306 tricks (10.2306), sets 467 times (4.67%) with imps of -7789 (-0.7789).

S6 lead yeilds 102268 tricks (10.2268), sets 508 times (5.08%) with imps of -7405 (-0.7405).

S2 lead yeilds 102266 tricks (10.2266), sets 510 times (5.1%) with imps of -7384 (-0.7384).

H8 lead yeilds 99766 tricks (9.9766), sets 753 times (7.53%) with imps of -2760 (-0.276).

H7 lead yeilds 99766 tricks (9.9766), sets 753 times (7.53%) with imps of -2760 (-0.276).

H3 lead yeilds 99866 tricks (9.9866), sets 739 times (7.39%) with imps of -2984 (-0.2984).

D9 lead yeilds 99261 tricks (9.9261), sets 1011 times (10.11%) with imps of 0 (0.0).

D8 lead yeilds 99261 tricks (9.9261), sets 1011 times (10.11%) with imps of 0 (0.0).

CJ lead yeilds 101494 tricks (10.1494), sets 590 times (5.8999999999999995%) with imps of -5881 (-0.5881).

C7 lead yeilds 100492 tricks (10.0492), sets 719 times (7.19%) with imps of -3788 (-0.3788).

C4 lead yeilds 100493 tricks (10.0493), sets 722 times (7.22%) with imps of -3760 (-0.376).

The panel got this right :)

[Mbodell]

After rechecking the official leading agreements, I led ♥7, which was worth zero points :( This was the second consecutive MSC quiz, where the "top-of-nothing" false-card scored more than the systemic "second-highest" lead :( The panel got this right :)

 

I think the "second highest from weak suits" is talking about from 4+ length suits. From 1 card, you obviously don't lead second-highest. From 2 cards, you also shouldn't lead second-highest (lead highest). From 3 cards, you also shouldn't lead second-highest (lead low from honor or highest from nothing). At least that's how I'd read it.

[the_clown]

I would lead a ♥.

 

If partner was desperate for a diamond lead he may have well risked a 2♦ overcall (he is NV). Clearly he will need an excellent 5 card suit or 6 card suit for an overcall. Leading a club is out when partner didnt double 2♣, besides Jxx is a terible holding to lead from.

 

Leading a spade will often cost a trick and even when partner has an honour, we have no entry to cash the spades.