[Siegmund]

I know the textbook answer -- "you need at least QJ9 to lead a sequence... you're trying to set up the suit against notrump, not just one trick like you are in a suit", you don't want to burn 2 honours on one trick if partner is short, you want to keep a high card to get back in to run the suit, blah blah...

Problem is, I've been spending some time fussing with double dummy simulations lately, and it looks like that isn't how it works. "Lead x from QJxx at trick 1, and thereafter defend double-dummy" averages 0.07 tricks worse than "lead Q from QJxx at trick 1 and thereafter defend double-dummy", over a large sample of 1NT-3NT hands. (

I inspected several dozen of the deals where Q and x lead to different numbers of tricks to see what the common features were. The results were mostly what you'd expect: x gains when partner has a doubleton ten or king, or dummy has a singleton ace or king. Q gains against a lot of positions where declarer has AKT (or AT/KT in his hand and the other honor in the dummy, and partner has the nine), and when partner has K9x(x) and we need to preserve the tenace. I did not see very many hands where third hand actually needed to know whether opening leader had a 3-card sequence to know what the correct defence was -- the "partner needs to know how good your sequence is to decide if it's worth returning the suit" argument doesn't appear to hold water.

That really only seems to leave one situation as a possible explanation for the discrepancy between the simulation and standard practice -- the ten (or ATx/KTx) is in the dummy, and double-dummy the best play is to rise with the ten at trick one rather than play second hand low, while real-life declarers are afraid to burn the ten, on the theory that Q9xx+J9xx is twice as likely as QJxx with opening leader.

Is that one situation enough to turn the balance against the honor lead? Has anyone actually tried systematically leading high from QJxx (and KQxx and JTxx) against notrump and found it works badly, or all we all just blindly following some ancient advice that turns out to not be quite right?

[babalu1997]

well, i think there are a couple of examples in william root`s books advising the q lead if dummy has shown a 4 card suit.

there maybe a couple more examples there if you care to look.

[jdonn]

Lead low, Because even when queen is "right" or when they are equal leading low works better single dummy. Some examples.

---

Leading low costs a trick double dummy but not in practice since declarer plays the king.

---

It looks like they are equal double dummy but leading low is better, similar to above.

---

This is the type of situation you mentioned. Equal double dummy, but low better in practice.

---

Low is a trick worse double dummy. In practice on the queen lead declarer can win in hand and lose a trick to partner who will just give the trick back anyway.

[Siegmund]

Thanks for pointing out some further instances where the single and double dummy plays differ. I can believe there are enough cases to make x better now. (And I now have some ideas where to look when I look at the KQxx cases in more detail. The difference there is a lot bigger - 0.2 tricks - so I will be needing quite some number of reasons for x )

[awm]

Another observation is that the spot cards matter. For example, I've found that leading the queen from QJ87 or king from KQ98 often works out well despite being one pip below the normal expectation, and I do make honor leads from these holdings sometimes (more frequently when leading from a five-card suit though). On the other hand, queen from QJ32 seems like a losing strategy.

[MFA]

awm, on Feb 3 2010, 12:25 AM, said:

Another observation is that the spot cards matter. For example, I've found that leading the queen from QJ87 or king from KQ98 often works out well despite being one pip below the normal expectation, and I do make honor leads from these holdings sometimes (more frequently when leading from a five-card suit though). On the other hand, queen from QJ32 seems like a losing strategy.

Right, and the bidding matters also.

1NT-3NT is different from 2NT-3NT for instance, since the odds of finding declarer with A+K are much different.

I think leading low should be default but that one should judge from time to time when holding these 2-card sequences. I have observed that I tend to lead an honour a little more often than most good players in these situations, but I think I have been reasonably successful doing so, and I only do it when I think the situation specifically justifies it.

[mich-b]

MFA, on Feb 2 2010, 06:23 PM, said:

I think leading low should be default but that one should judge from time to time when holding these 2-card sequences.

What about leading from KQxx ?
is your "default" still the "x"?