Saturday, October 16, 2010

Rosenblum trump compound

[Andy]  It's my turn to blog on Franco's favorite subject.  The following is what will most likely be the most famous hand from the Rosenblum final.  (It was described in the NY Times today also, although I didn't see it online there yet.)

♠ A 6 3
K J 
A K 4 2
♣ K 4 3 2
♠ Q J 4 2♠ 10 9 7 5
8 3Q 9 2
Q 8 7 6 5♦ 10 3
♣ 9 8♣ Q J 10 6
♠ K 8
A 10 7 6 5 4
J 9
♣ A 7 5

Zia got to 6 and received the ♠Q lead.  He found an elegant line, cashing his winners and taking 3 ruffs in his hand, then exiting with his last club.  In the end position he could claim regardless of the location of the Q.

Moss and Gitelman did one better, though, arriving in 7.  I'm not sure what happened, unless Gitelman intentionally bid a grand on a blackwood auction in hopes that the opening lead would reveal the trump Q.  (This is a somewhat well-known trick, but definitely pretty gutsy in the last quarter of a world championship final that you are winning!)  Anyway, Rodwell did lead a trump, and Moss won the trump A, played trump to K, spade to K, drew the last trump, ruffed the third round of diamonds, played the second to last trump, unblocked clubs, and played the last trump to achieve a double squeeze (spades being the double threat suit).

That was very well done.  Let's take a look at the hand from a theoretical point of view.

With a reasonable number of entries, compound squeeze principles suggest that the contract should be makeable if one positional threat can be found.  It seems natural to hope that suit is diamonds, since declarer can isolate it, and the club threat needs to be in his hand.  Moss' line probably would not have worked had Rodwell held something like ♠J x x  x x  Q x x x x ♣ Q 10 x.  Rodwell would probably unguard spades and keep clubs in that layout, and Moss would not have the timing to unblock the ♠A to achieve a double squeeze around clubs.  (This is necessary since the extra spade winner gives Rodwell an idle card.)

So to achieve full compound generality, declarer appears to need to preserve the ♠K in his hand.  Perhaps declarer could take the diamond ruff before drawing the last trump.  This would leave two entries in both black suits and guarantee a working compound position as long as Rodwell held at least 4 diamonds.  Of course, it might lose if diamonds were 6-1 and RHO could ruff.  It might also lose if LHO overruffs in a position where there was just a minor suit squeeze against RHO the whole time.

An intriguing variant is to start the way Moss did, but play the 4th trump before taking the diamond ruff.  I think that also guarantees the ability to make the contract any time LHO has at least 4 diamonds.  The triple squeeze part of the compound is achieved at that point, since there is a ruff-out threat in diamonds.  Then, if LHO unguards spades at that point and declarer can read it, he can unblock the ♠A before taking the diamond ruff.  If LHO guards spades then the play continues as it actually did at the table.  Cool! Read more!

Sunday, October 3, 2010

Another Compound

Dlr E
Vul NS

♠ A 8 7 3
A Q 10 2
K 7 6 5
♣ 6
♠ K 2
K J 8 7
♣ A K Q 10 4

3♣ 3NT
5♣ 6♣ Pass 7NT
All Pass

♣3 led. RHO puts up the Jack. How do you play?

FWIW, RHO is an honest customer, LHO not so much. You can confirm this by playing another club.

My line and further analysis is below.

I tested one more club and LHO pitched ♠4 (I pitched ♠3 from dummy). Then 4 rounds of hearts, LHO following up the line, while RHO pitched ♣7, ♠5, then 4.

There's definitely a squeeze available at this point, but I still haven't committed to which one to play for. ♠K drew the Jack from RHO and 6 from LHO. Now I can afford another club before making up my mind: LHO pitched ♠9 while RHO followed with the 5. This is the ending:

♠ A 8
K 7 6
♠ 2
♣ 10 4

RHO is down to a club stopper and 3 pointed cards. If he started with 4 spades (and so still has that suit stopped), LHO made a very tricky pitch from 3=4=5=1 at trick 2 (and again on the last trick). Not to mention that the aggressive 3♣ opening seems much more likely with 4 diamonds than 4 spades. So, I played a spade to the Ace, and when RHO showed out I claimed: he must pitch his diamond guard here, then A, ♣10 squeezes LHO in diamonds and spades.

If I had thought that RHO guarded spades but not diamonds, then K, A, ♣10 would effect a similar squeeze.

Technically this was just a double squeeze (LHO guards spades, RHO guards clubs, both guard diamonds), but if RHO had started 3=1=3=6 it would have been a true compound (a Type-R Unrestricted).

How about the defense?

East is going to need to make 2 pitches on the hearts, and then declarer is going to need to guess which of the pointed suits East still guards. If, as on the actual hand, East pitches one from each suit, he'd better not have started 3=3 or declarer cannot go wrong. On the other hand, even though East has had to make 2 pitches, always pitching one from each reveals nothing about which suit started longer. If West also pitches one from each suit (he has to make 2 pitches on clubs), declarer faces a straight guess (ignoring the auction), even though this is (or at least looks like) an unrestricted compound.

In compensation, if East did start 3=3, he must pitch 2 from a single suit, and declarer will almost always go right by playing that suit to now be unguarded. When East is actually pitching 2 from a 5 card suit (which he should do instead of 1 from each), declarer should normally go wrong (3-3 vs 5-1 is 25:3 a priori, though knowing an opponent's preempting style might be enough to overcome that).

I guess when the layout is such that it's not a true compound, just a double, that the position-tightening does not actually put anyone under pressure and so it makes sense that proper discarding reveals virtually nothing. There's always a double squeeze present, but it's more or less a toss up as to which to play for if both defenders are perfect. In contrast, when it is a true compound, East is squeezed twice and his discards are much more revealing, making it around 80% that declarer gets it right even against perfect defenders.

Obviously in practice a psuedo-squeeze is much more likely than this layout and West should be minimizing the chance of screwing that up.

Some minor technical points:

1. I should not have played a club at trick 2. RHO might not appreciate what is going on and if he pitches 2 clubs I'm spared any guess.

2. If I do play a club, I might as well pitch a heart. LHO might have trouble pitching from a 4 card holding in diamonds or spades. Also, I might as well play a 3rd (but not a 4th!). This gets more information from LHO early, without really helping RHO much if at all.

3. When crossing back to hand, I should probably use A in case the Q has been bared.

Finally, the J was not strictly necessary, but it does help slightly in constraining the defense. Earlier I suggested that LHO pitching a spade early (and another later) from 3=4=5=1 would be tricky. This follows the general "principle" (technically this principle is false, but that's another post) of not waiting to the last possible moment to release a guard if you're squeeze. Since West is going to be squeezed in diamonds and spades later in the hand, if he anticipates this he has a good chance of leading declarer astray by pitching from his shorter suit early. However, thanks to the J, he actually can't get away with pitching 2 diamonds (he started with Qxx). At double dummy the clash menace is irrelevant, but at single dummy it puts additional constraints on the defense and may make reading the position slightly easier.

The BBO record is here.

Read more!