[hrothgar]

AKT98x would have been a lot easier...

I'm going to lead low here (I would have lead the 10 off the first)

[mikeh]

I've always led low from these holdings, but (even tho I will still lead low here) I am increasingly finding that I am losing imps by doing so: lost a game swing in Verona for example... the opps led high and ran the suit. Was that just bad luck? I like to think so

[Winstonm]

Isn't the only relevant holdings 322 - if one of the opps held a singleton spade they might have looked for a suit contract, and if partner is stiff then the suit won't run and I'm entryless - so it seems that low should be best as I'm bucking 2/1 odds with 1 partner and 2 opponents that my partner has the 3-card holding.

The other question is do they have 8 fast tricks outside my suit? The only time it is right to lead low is when partner has xx (forgetting Qx for this debate) and an entry to stop their 8 trick run; but leading high only wins whenever partner is dealt xxx.

A last concern is when partner has Qx, or QJ and the high-card lead blocks the suit.

It seems not too close to me - low - when the suit is right to lead - has to be the favorite.

[the hog]

Lead low. Better than leading an hon, finding pd with xx in S and an entry and the suit distributed 3-2. Try explaining that to him and you won't get much sympathy

[cherdano]

Let's assume the suit is 6322.

Leading high wins immediately when partner has 3, which is happens 1/3 of the time, or when the 3-card suit only has low ones, which adds another 9%.

Leading low wins immediately when partner has the queen, which a priori happens about 1/3 of the time. Plus probably when partner has the Jack but not the ten, and LHO the queen, which happens 7.4% of the time according to my calculations.

So the chances to beat it directly are pretty similar - except for the problem that partner is less likely to hold the queen than opps, who have some 26-31 hcp. So the problem is whether we make up for this deficit by the chance that partner has a doubleton and an entry before they take 8 tricks in the other 3 suits. E.g. if we think partner has the queen 20% of the time, then it would be enough if we beat it 25% of the time partner has a low doubleton.

It seems to me the best chances to beat it via the spade suit is a low one. How good is the additional chance to beat it by leading high and finding a shift? Pretty low for such a hcp auction I would think, but it might become more attractive when it is clear that they are bidding 3N on a running suit and little outside.

Arend

[the hog]

Jlall, on Oct 23 2006, 01:15 PM, said:

Winstonm, on Oct 22 2006, 09:15 PM, said:

so it seems that low should be best as I'm bucking 2/1 odds with 1 partner and 2 opponents that my partner has the 3-card holding.

There are some flaws in this argument...

First of all, there may be other defenses to beat the contract. Leading high and finding a killing shift is possible.

Secondly, if you can run the suit immediately, you will certainly beat the contract. However even if you catch partner with a doubleton (which you rate to do when the suit is 322 as you say) you still havent beaten it yet. A lot of the time the opponents will just run 9 tricks. Leading low even if you catch partner with a doubleton you still need him to have an entry and enough stoppers to stop the opponents from running the suit (unless his doubleton is Qx).

I would bet that you will not beat the contract half the time that you find the suit 322 with partner having 2 anyways which would mean leading high is more likely to beat the contract.

Hmm. Interesting.
Ben, would you be prepared to do some of your thorough research on this? I, for one, would be very interested in the results.

[cherdano]

hrothgar, on Oct 23 2006, 01:58 AM, said:

AKT98x  would have been a lot easier...

I'm going to lead low here (I would have lead the 10 off the first)

I don't understand your reasoning here. Leading high from AKT98x works more often, as it is more likely that the 3-card suit can't stop us at trick 3. (Assuming 6322 split, the total a priori chances are 51% instead of 42%.)

Arend

[Flame]

I dont know the answer (i tend to lead small) but i think there are factors that matter. First is entry to your hand, if you think you got an entry like having something good in a short suit then leading high is more likely to be better.
Second is whather its a major or minor suit, after 1NT-3NT its more likely to find the opponenets 2=2 in a major then in a minor, so a major high is more likely to work then a minor high.

[Echognome]

I have given this problem some thought theoretically and my conclusion is that it's very complicated.

We have two major unknown subjective probabilities to deal with. First off is the probability that if we lead high and we need to switch that we find it. The second probability that is unknown is if we blow a trick at trick 1 that they do not have 8 other tricks. That is to say, we may be giving them their 9th trick at trick 1.

Whatever weights you give to these unknown probabilities will have a big effect on any conclusions we draw.

Looking at some specific holdings around the spade suit when we lead low:

6=2=3=2 (that is 6 for opening leader, 2 for dummy, 3 for partner, and 2 for declarer). If say that an upper bound for the location of the Q in partner's hand are the vacant spaces (I believe it's less), then 4/7 of the time we are blowing a trick at trick one. And probably getting a surprised declarer when he wins the first trick with Qx opposite Jx or xx.

6=2=2=3. In 2/7 of the cases, we set outright. (again an upper bound) In the other 5/7 of the cases, we only set when they don't have 8 other tricks. We also may have blown a trick when partner has JT or when partner has Jx and declarer has to guess, albeit this is unlikely. However, if there is Qx in dummy and Txx in declarer's hand we may set a cold contract.

6=3=2=2. In 2/7 of the cases, we set outright. In the other cases we may put declarer to a guess (Qxx opposite Tx) and we may cost a trick outright (xxx opposite Qx).

6=3=1=3. In almost no case are we going to set. But we may be giving up a tempo.

6=2=1=4. We may be blowing a full trick here. In other cases, we are almost surely losing a tempo.

Then again, when we lead high, we have to figure out whether to continue or switch and that seems like it could be complicated as well.

I guess what I'm saying is that theory isn't going to take us very far. I imagine that simulation is the only real way to answer this question.

[Halo]

It is true that opponents can have exactly eight tricks and partner have three solid tricks. But partner usually also needs a lower missing honour to stop oppos establishing a trick by force in the suit you led, and even that only helps half the time when the queen is under the AK. Of course partner may have three fast tricks, but that has to be unlikely on about 7-9 points

It feels much more probable that opponents don't have eight quick tricks and have to concede one trick to partner.

I would therefore lead small.

[FrancesHinden]

I'm generally a low leader. To compare with mikeh's loss in Verona, I won a big swing recently from leading low when partner had Qx.

We are looking only at the case where we don't have an outside entry, as of course we would have bid with an outside ace or king.

I believe the chance of the suit being 6322 is lower than most people seem to be suggesting. Holding something like

x
Ax(x)
xxx(x)
AQ10xxx(x)

most people just raise 1NT to 3NT - particularly if to show a club suit they would have to bid 2S.

Of course, a related upside to leading a top one is one dummy has a singleton honour and partner 10x.

[kgr]

In Jack after 1NT-3NT with W:
AK9875
xxx
xx
xx
=> 1000 samples:
SA: -327.9
SK: -352.9 (no idea why, maybe unblocking is expected/not expected)
S5: -364.1
S9,8,7: -388.2
Hx: -494

[FrancesHinden]

You'll have to specify the 3NT bid fairly carefully, not only to include all the balanced hands while excluding hands that would have looked for a heart fit; but to include unbalanced hands with a long minor that are both strong enough to want to punt 3NT but not strong enough to look for alternative contracts. And the exact specification of those hands depends on the opponents' methods.

(For example, I play 1NT - 2S as a xfer to clubs, and 2NT as a xfer to diamonds. Opener is supposed to complete with Axx, Kxx or better. So you would make the xfer on a hand where that holding opposite tells you if the suit is running e.g. AQxxxx and, but you wouldn't bother xferring if that isn't what you are after (e.g. AKxxxx you just guess if the suit is running or not). I'm not trying to claim these are the best methods, but they affect what hands raise directly to 3NT. Also, because we use 2S as a club transfer, we aren't going to bother transferring on a marginal raise to 3NT with spade shortage to avoid getting doubled)

[cherdano]

With 6331 shape and strong enough to bid 3N, I would usually transfer to my minor and then show my singleton at the 3-level. With 13(54) I would bid 3M. Both are not necessarily slam tries, but can just be choice of games. Is this non-standard?

I would also like to know the exact specification of the Jack analysis. Was it double dummy? What is Jack's style for 1N-3N?

Arend

[gwnn]

4th best (no entry)

[kenrexford]

I'd overcall 2♠ and save myself the grief.