A definite bug in GIB declarer play

[dartagnan9]

Hi, if you could attach screen captures, that would be a big help, since this bug is obvious. Or if you could at least use some BBO function to reference a game that other people could then see.

 

North declarer in 3NT (bidding South 1♠, North 1NT, South 2 ♦, North 2NT, South 3NT). Bidding gave no information about cards in opps' hands.

 

1 round of clubs is played. East opens the lead with the J clubs, which clearly implies he has the 10 clubs too. East won with the Ace, leaving the following:

 

North (Declarer) K,Q,9,3

East 10,7

South 6

West 4,5

 

East leads 5 clubs, North plays 9. hoping to finesse 10, East takes it with 10, last club removed from dummy.

 

What are the odds that's the correct play? They're very easy to calculate. The possible distribution, after 5 is led, is

 

West East

0 3 - this is the only scenario where the finesse needs to be played

1 2

2 1

3 0

 

So the odds of that being the right play are 25%.

 

Now, if East had J 10 xx before his initial lead, then the finesse is going to fail anyway so might as well not try it since the odds of it being the right play are only 25%. Moreover, the opening lead clearly indicated the 10 in East's hand, in which case, if he didn't have the J, then not trying the finesse is the correct play.

 

It sounds complicated, but it was blindingly obvious to see the right play, and exceptionally infuriating to see GIB make the unnecessary finesse. Now if I can calculate the odds so easily, and illustrate the logic behind deciding where the 10 clubs is likely to be, why can't GIB?

 

This is not the first time I've seen unnecessary finesses. I think there's something wrong in the algorithm which triggers trying a finesse.

 

Thanks, Keith

P.s. I have the jpg, if there's a way to post it.

[ahydra]

Did you mean West wins with the Ace and West plays the ♣5?

 

The odds of a 3:1 or 2:2 break remaining (I include the ♣5) are NOT equal. For a 3:1 break remaining there must have been a 4:2 break to start with so that's 50%, similarly 2:2 came from 3:3 which is 36% a priori. If we ignore 5-1 breaks (West holding 5) the odds are actually approximately 58% and 42% (50/86 and 36/86).

 

The finesse should not be used when they are breaking 2:2 with East holding the 10 - but if East led from Jx, then the finesse gains a trick. As you said, if East holds J10xx nothing can be done. Hence GIB correctly takes the finesse, with rough odds of 79% of coming out at least equal or ahead of playing K Q. (58% for the 4-2 breaks: if West has 4 you win and if East has you're stuck anyway; and 21% for the 3-3 breaks with West holding the 10).*

 

Even if you assume East has the 10, you still have 58% odds of coming out equal or ahead by playing the finesse (we lose the positions where West has 3 cards including the 10).

 

There are situations where it's better to play the K Q though - if North has only one entry, or can't afford to lose too many tricks, for instance. If the 10 hasn't dropped, he can play a fourth round while still maintaining the entry. So without seeing the hand it's impossible to tell whether GIB is playing correctly or not.

 

For the image I suggest you use imageshack or some other image hosting site and post the link here. Next time though there's a handy tool (if someone else can be kind enough to tell you about it, as I don't know how it works? :ph34r:) where you can link directly to the hand in question.

 

ahydra

 

*The odds are slightly less than this because we didn't take into account East holding J10 doubleton.

[Bbradley62]

Dartagnan9: Instead of a screen capture, you can provide a link. This also allows the viewer can even step through trick-by-trick.

 

If you're looking at the hand in "My Hands", there's an "Export deal" button at the bottom; click on it and select "Handviewer link". This will create http:/tinyurl.com/something, and you can copy that link into a forum thread.

[dartagnan9]

That's great to know about the viewer link, thanks!

 

Maybe it's because it's the end of a long Saturday, but I can't follow the odds calculations, but I'm sure you're right, ahydra - statistics & probability were never my strong suit (no pun intended)!

 

Just one point though, and then I'll shut up. It may be that this is the human judgment that comes in: it seems unlikely to me that East would have led the jack without the 10. It's impossible to calculate those odds because it depends on your assessments of how many people would play the jack in that situation lacking the 10. My assessment is that few would, but I'm almost certainly not as experienced in the science of opening leads as many on the board, so I grant I could be wrong.

 

But if you take my assessment as correct, then the only time it's worth taking the finesse is if East has J 10 only. And to me it seems vastly unlikely somebody would lead from J 10 doubleton against NT.

 

Of course, if I'd included the link, it would have been clearer since you could judge whether the distribution in the other suits precluded not leading clubs in the first place.

 

I suppose this is what makes bridge so interesting! Thanks Keith

[barmar]

If you think this was bad, see this thread:

 

http://forums.bridgebase.com/index.php?showtopic=41296