[ceeb]

♥J led, Q, K

---

NS are U.S. players in IMP quarterfinal at US nationals; I guess they are playing 2/1 style. EW are Zia, Hamman in case it matters.

[dake50]

I bite.
DQAK dis Hx. CAKQ dis H. Run SJ.

[nige1]

ceeb, on 2010-November-29, 20:20, said:

♥J led, Q, K
NS are U.S. players in IMP quarterfinal at US nationals;
I guess they are playing 2/1 style. EW are Zia, Hamman in case it matters.

---

Dake50's line has about the same chances but I slightly prefer ♥A, ♣K, run ♠J and repeat the finesse if successful.
Against lesser players, if ♠J is not covered, then you might follow Zia's tip and switch horses: rise with ♠A to cash ♣A, ♦Q, ♦A, ♣Q, ♦K.

[ceeb]

If you go for immediate discards, dake hit the point that ♦ first is right because you have a chance to recover if RHO ruffs the 3rd one. However, on simple % calculation I think the trump finesse is a few % better. The imponderable question on which I hoped for insight is whether, in view of the small difference in the various probabilities, that even though Zia might be better than most at ducking the ♠J, he might still err often enough to make it worthwhile to test him then switch horses if he does not cover the trump J. Most players can't hope to outclass Hamman and Zia, but still hope to beat them somehow. Hence you hope that they make mistakes than you. Is the best strategy just to hope by chance that the cards deal them more opportunities for wrong decisions than they deal you? Or should you actively push them when the opportunity arises?

In practice declarer Siebert (I forget which brother) ran the trump J and went down. All suits split.

[gnasher]

Of the two suggested lines:

- Spade finesse = almost 50%

- Diamonds, then clubs
= 36% (diamonds break) x 85% (clubs not 6-2, adjusted for the possibility of a ruff from Kx) = 31%
+ 13% (diamonds 4=2 with ♠K onside, adjusted for vacant places)
= 44%

So not that close when you're playing against demigods.

Would Zia cover from something like Kx Kxx Jxxx Jxxx? Probably not. It would be right if declarer had AQ8xxx Axx xx Ax, but that would mean declarer had grossly overbid

By playing ♠J to the ace, you lose the chance of ♠Kxx and ♦xx onside, so this line is only about 40% a priori. Therefore you need Zia to not cover about 1/5 of the time. That seems unlikely.

This post has been edited by gnasher: 2010-December-01, 05:31

[Flameous]

Note that you also win the slim chance of singleton king behind.

[OleBerg]

If I could think this one up relatively fast at the table, it would definitely be my plan:

After ♥A, three rounds of clubs, discarding the ♥3. Now the ♠J.

Now, to East it migth look like the only hope to defeat it, is to get a spade and a diamond, making it much more attractive to cover with the King. (I rise if not covered, playing on diamonds.)

I believe this would bring my percentages close to the 50% of the finesse.

Furthermore, anytime my plan succeeds, I have won a small psychological victory. They will know that they can't nescesarily know excactly what to expect.*

Thirdly; the joy of beating Zia at his own game.


'Ok, against me, they will probably figure that out fast anyway. but probably not in a way that makes them feel uneasy.

[ceeb]

gnasher, on 2010-December-01, 05:24, said:

Of the two suggested lines:

- Spade finesse = almost 50%

Thanks for the analysis. Definitely helpful. Yes, I underestimated the ♠ finesse line by underrating its chances against ♠Kxxx(x). It seems to be about 48%.

Quote

- Diamonds, then clubs
= 36% (diamonds break) x 85% (clubs not 6-2, adjusted for the possibility of a ruff from Kx) = 31%
+ 13% (diamonds 4=2 with ♠K onside, adjusted for vacant places)
= 44%

Those figures even seem to me to be generous, especially the 13% which neglects that ♣ must still break. Overall I get only 40%.

Quote

So not that close when you're playing against demigods.

There's about 32% when every line works. In the relevant sense that finessing picks up twice as much of the remainder as discarding it does seem not very close.

Quote

Would Zia cover from something like Kx Kxx Jxxx Jxxx? Probably not. It would be right if declarer had AQ8xxx Axx xx Ax, but that would mean declarer had grossly overbid

To be fair, whether Zia would cover doesn't depend merely on whether it's right or wrong on analysis. Surely he will only do as much thinking as he has time for in advance, so if we play like Hoffman we get more edge. Second, there is at least a slight chance that even if Zia doesn't see any advantage to covering he also won't clearly see the advantage to ducking, and on some of those occasions of ignorance he'll go wrong under the time pressure of second hand play.

Quote

By playing ♠J to the ace, you lose the chance of ♠Kxx and ♦xx onside, so this line is only about 40% a priori. Therefore you need Zia to not cover about 1/5 of the time. That seems unlikely.

I don't see offhand how to estimate the necessary chance for Zia to cover from the foregoing figures, and the wording is confusing as well.

[gnasher]

ceeb, on 2010-December-01, 13:31, said:

To be fair, whether Zia would cover doesn't depend merely on whether it's right or wrong on analysis. Surely he will only do as much thinking as he has time for in advance, so if we play like Hoffman we get more edge. Second, there is at least a slight chance that even if Zia doesn't see any advantage to covering he also won't clearly see the advantage to ducking, and on some of those occasions of ignorance he'll go wrong under the time pressure of second hand play.

I'd be surprised if Zia allowed himself to be talked into playing more quickly than he wants to.

Quote

I don't see offhand how to estimate the necessary chance for Zia to cover from the foregoing figures, and the wording is confusing as well.

I think my arithmetic was wrong, but this is how to compute the correct answer:

Suppose that against perfect defence the chances of success are:
(1) Spade finesse: 50%
(2) ♠J to the ace, then cash diamonds: 40%

The practical chance that line (2) works is

40 + 60 / 2 x P

Where P is the probability that RHO will cover when he shouldn't (divided by 2 because he has to be dealt the king in order to misdefend).

So the break even point is

40 + 60 / 2 x P = 50

Or

P = 1/3

[ceeb]

Of course (hits head). One could expand the model putting f=probability the finesse line will win and writing Pf instead of P/2. Then the condition for playing the fake finesse becomes approximately
0.4 + 0.6 Pf > f, or
0.4 > f(1-0.6P), satisfied either by large enough P or small enough f -- in particular f<0.4 no matter what P is.

At this point, having discussed the hand, it is easy to grasp why RHO should not cover the ♠J and therefore to suppose it is obvious and no top player would go wrong. But if we allow that even Zia would take a moment to work it out and imagine that at trick 1 we manage to cover the opening lead within a couple of seconds and Zia plays along in tempo, and then at trick 2 as we cross in clubs he follows suit without betraying any wistful delay as if getting ready for the next trick, and that at trick 3 when the ♠J is led he of course follows low in tempo, then we have evidence for the smallness of f (evidence that he doesn't have the ♠K), a good case for going up with the ♠A.