GIB doesn't give it a chance?

[Bbradley62]

At Trick 3, it appears to me that North's only chance of making the contract is to cash ♥A hoping to drop the K, enter hand with ♥9 and throw ♣10 on ♦A. GIB doesn't run enough simulations to have that possibility come up?

 

[hv=lin=pn|puerch,~~M21444,~~M21442,~~M21443|st%7C%7Cmd%7C4S458QKH3TJQADC6TA%2CS27H5D23579TKC28Q%2CS69H24679DJQAC459%2C%7Crh%7C%7Cah%7CBoard%202%7Csv%7Cn%7Cmb%7C1C%7Can%7CMinor%20suit%20opening%20--%203%2B%20C%3B%2011-21%20HCP%3B%2012-22%20total%20points%20%7Cmb%7C2C%7Can%7CMichaels%20--%205%2B%20H%3B%205%2B%20S%3B%209%2B%20total%20points%3B%20forcing%7Cmb%7C3C%7Can%7CCompetitive%20raise%20--%203%2B%20C%3B%207-9%20total%20points%20%7Cmb%7C4H%7Can%7C5%2B%20H%3B%2012-%20total%20points%20%7Cmb%7Cp%7Cmb%7C5H%7Can%7C2%2B%20C%3B%205%2B%20H%3B%205%2B%20S%3B%203-%208421%20HCP%20in%20C%3B%2033%2B%20total%20points%7Cmb%7Cp%7Cmb%7Cp%7Cmb%7Cp%7Cpc%7CC3%7Cpc%7CC6%7Cpc%7CCQ%7Cpc%7CC4%7Cpc%7CC8%7Cpc%7CC5%7Cpc%7CCK%7Cpc%7CCA%7Cpc%7CCT%7Cpc%7CC2%7Cpc%7CC9%7Cpc%7CCJ%7Cpc%7CH8%7Cpc%7CH3%7Cpc%7CH5%7Cpc%7CH9%7Cpc%7CDA%7Cpc%7CD4%7Cpc%7CS4%7Cpc%7CD5%7Cpc%7CH6%7Cpc%7CHK%7Cpc%7CHA%7Cpc%7CD2%7Cpc%7CHQ%7Cpc%7CD3%7Cpc%7CH7%7Cpc%7CS3%7Cpc%7CSK%7Cpc%7CS2%7Cpc%7CS6%7Cpc%7CSA%7Cpc%7CD6%7Cpc%7CHJ%7Cpc%7CD9%7Cpc%7CDJ%7Cpc%7CSQ%7Cpc%7CS7%7Cpc%7CS9%7Cpc%7CST%7Cpc%7CS8%7Cpc%7CDT%7Cpc%7CH4%7Cpc%7CSJ%7Cpc%7CDQ%7Cpc%7CD8%7Cpc%7CHT%7Cpc%7CDK%7Cpc%7CS5%7Cpc%7CD7%7Cpc%7CH2%7Cpc%7CC7%7C]400|300[/hv]

[pigpenz]

yes it does appear that way

[barmar]

I found online copy of Ginsberg's article that describes how GIB works.

 

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.52.2188&rep=rep1&type=pdf

 

GIB doesn't try to find a hand that allows the contract to be made. It deals a bunch of hands consistent with the bidding and play, and finds the best double-dummy score of each legal next play. Then it chooses the play whose average score across all the deals is highest. It's looking at the AVERAGE result, not the best possible result of each play.

 

This explains why GIB doesn't take low-percentage plays, even if they're the only way to make the contract. In a vulnerable game, a play has to have at least 15% of succeeding, versus going down 2, to make it choose it over a play that holds to down 1. For a non-vul game, it has to be 20%.

 

Since West raised clubs, East can have at most 4 clubs. It also has at most 4 spades, since it didn't open them. But it can't have 4=1=4=4 shape, because it would open 1♦ with that. So East has to have a weak NT hand, with either 2 or 3 hearts. Cashing the ace only works if West has the singleton K, but since East opened, it's more likely to have the K.

 

This all adds up to there being much less than 15% chance of making the contract, so it goes for the almost sure down 1 (in fact, the only layout where it goes down 2 with this line is the one where it could have made with your line).

 

When I first was writing this response, I included a caveat about East overruffing when it tries to get to the board with a spade ruff for a ♥ finesse. But that can't happen, because East can't have only 2 spades. Its shape would then be either 2=3=4=4 or 2=2=5=4, both of which would open 1♦.

[xxhong]

A simple EV calculation:

the line of cashing HA:

5H making; 650, chance x (You drop HK)

5H down 2: -200, chance 1-x (You don't)

 

5H -1: -100, when HK finesse is on. (roughly 50%, can be slightly higher due to the bidding.)

5H -2: -200, when H K finesse is off.

 

If we take the percentage that HK is on as 50%,

 

650 x - 200 (1-x) = -100 * 0.5 -200 * 0.5

x = 0.05088

 

If we take the percentage that east holds HK as high 75%.

x = 0.08823

 

So it's safe to assume that if it makes 5H about 10 % of times, it should choose the line to drop in the risk of down 2.

If we estimate the original chance to drop HK, it is about 26%.

 

SO really, the computer makes a huge mistake here for not dropping stiff HK.

 

I found online copy of Ginsberg's article that describes how GIB works.

 

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.52.2188&rep=rep1&type=pdf

 

GIB doesn't try to find a hand that allows the contract to be made. It deals a bunch of hands consistent with the bidding and play, and finds the best double-dummy score of each legal next play. Then it chooses the play whose average score across all the deals is highest. It's looking at the AVERAGE result, not the best possible result of each play.

 

This explains why GIB doesn't take low-percentage plays, even if they're the only way to make the contract. In a vulnerable game, a play has to have at least 15% of succeeding, versus going down 2, to make it choose it over a play that holds to down 1. For a non-vul game, it has to be 20%.

 

Since West raised clubs, East can have at most 4 clubs. It also has at most 4 spades, since it didn't open them. But it can't have 4=1=4=4 shape, because it would open 1♦ with that. So East has to have a weak NT hand, with either 2 or 3 hearts. Cashing the ace only works if West has the singleton K, but since East opened, it's more likely to have the K.

 

This all adds up to there being much less than 15% chance of making the contract, so it goes for the almost sure down 1 (in fact, the only layout where it goes down 2 with this line is the one where it could have made with your line).

 

When I first was writing this response, I included a caveat about East overruffing when it tries to get to the board with a spade ruff for a ♥ finesse. But that can't happen, because East can't have only 2 spades. Its shape would then be either 2=3=4=4 or 2=2=5=4, both of which would open 1♦.