[wbartley]

Just play bridge robot.

---

Human is South. GIB transfers to clubs when it should just bid 3NT

[virgosrock]

wbartley, on 2017-December-11, 11:04, said:

Just play bridge robot.

---

Human is South. GIB transfers to clubs when it should just bid 3NT

GIBBO bidding with 8 HCP opposite 1NT is quite questionable mostly. Here it is stuck. Rule says MUST xfer with 6 clubs. Next rule says no stiff so cant bid a suit. Hoisted on its own petard.

vrock

[Stephen Tu]

I don't know that this one is so clear, really, playing a system with no club invite bid. There are a ton of hands you could have where game has no play, or just goes down. My old PC version of Gib seems to want to play in 1nt/3c nv at imps but bid the game if vul at imps which may very well be the best strategy. It rates blasting 3nt as sub-par at this vul.

There are a ton of way more ridiculous bugs to complain about before we should start nitpicking close judgment calls like this one.

[wbartley]

Stephen Tu, on 2017-December-11, 13:25, said:

I don't know that this one is so clear, really, playing a system with no club invite bid. There are a ton of hands you could have where game has no play, or just goes down. My old PC version of Gib seems to want to play in 1nt/3c nv at imps but bid the game if vul at imps which may very well be the best strategy. It rates blasting 3nt as sub-par at this vul.

There are a ton of way more ridiculous bugs to complain about before we should start nitpicking close judgment calls like this one.

There is no way that 3NT is less than 50% on average. Dummy's club holding will produce 6 tricks way more often than it won't.

This is not nitpicking. I predict that if you give this hand to 100 experts, 100 of them will bid 3NT regardless of colors and regardless of scoring method, assuming they aren't looking for a swing.

[virgosrock]

wbartley, on 2017-December-11, 13:57, said:

There is no way that 3NT is less than 50% on average. Dummy's club holding will produce 6 tricks way more often than it won't.

This is not nitpicking. I predict that if you give this hand to 100 experts, 100 of them will bid 3NT regardless of colors and regardless of scoring method, assuming they aren't looking for a swing.

Not that clear wbartley. 1/3 of the time pd will have CK and high probability you are cold. Similar when NT opener LHO has Kx. When not Kx then nasty guess on second round. RHO will always duck the first round. I think Stephen is right. This is a gambling bid to be ok when vul, not ok when non-vul. Experts will take other things into consideration. The other room, what the field will do etc. etc.

vrock

[smerriman]

wbartley, on 2017-December-11, 13:57, said:

There is no way that 3NT is less than 50% on average. Dummy's club holding will produce 6 tricks way more often than it won't.

A quick sim of 500 hands showed 3NT making 50.8% of the time based on double dummy analysis. But since we're comparing to 3♣ (which makes 92.6% of the time), rather than a partscore in no trumps, it's a lot closer than it sounds. Looks like 3NT comes out on top, but not by more than 1 IMP.

[Stephen Tu]

My sim agrees going for 3nt is best double dummy.

But still I feel it is close and there are far clearer fish to fry than nitpicking GIB's simulations here that signoff is better nv. I want programmers to concentrate on correcting the truly ridiculous stuff before they start tweaking fine judgment where at least it is thinking along the right lines and has reasonable calls to choose from.

[steve2005]

wbartley, on 2017-December-11, 13:57, said:

There is no way that 3NT is less than 50% on average. Dummy's club holding will produce 6 tricks way more often than it won't.

An expert defender will duck with Kx offside and you are probably down when double dummy makes.

[wbartley]

virgosrock, on 2017-December-11, 14:06, said:

Not that clear wbartley. 1/3 of the time pd will have CK and high probability you are cold. Similar when NT opener LHO has Kx. When not Kx then nasty guess on second round. RHO will always duck the first round. I think Stephen is right. This is a gambling bid to be ok when vul, not ok when non-vul. Experts will take other things into consideration. The other room, what the field will do etc. etc.

vrock

You partner holds at least 15 HCP, you have 8. That means your partner holds at least 15 of the outstanding 32 points. Combine that with the fact that he has way more than two clubs on average and he's going to hold the club K about half the time, not 1/3 of the time.

[wbartley]

steve2005, on 2017-December-11, 15:27, said:

An expert defender will duck with Kx offside and you are probably down when double dummy makes.

That's small (doubleton king offside and declarer holding 2 clubs also) compared to the fact that double dummy opponents always make the right lead.

I would suspect that double dummy analysis is pessimistic from declarer's point of view. The number of situations where correct play and double dummy play by declarer changes the result is almost certainly less than the number of times the correct (or ambiguously correct) single dummy lead hands declarer his contract.

[virgosrock]

smerriman, on 2017-December-11, 14:15, said:

A quick sim of 500 hands showed 3NT making 50.8% of the time based on double dummy analysis. But since we're comparing to 3♣ (which makes 92.6% of the time), rather than a partscore in no trumps, it's a lot closer than it sounds. Looks like 3NT comes out on top, but not by more than 1 IMP.

smerriman does your sim include RHO ducking with Kx?

vrock

[virgosrock]

wbartley, on 2017-December-11, 15:54, said:

You partner holds at least 15 HCP, you have 8. That means your partner holds at least 15 of the outstanding 32 points. Combine that with the fact that he has way more than two clubs on average and he's going to hold the club K about half the time, not 1/3 of the time.

agree wbartley. rope in if pard has cK then a shortage somewhere else where they might run tricks the net effect may be slightly less chance. I go with smerriman's simulations where it is about 50% chance.

vrock

[wbartley]

In case anyone cares, a bridgewinners poll with 18 responders (for the moment) is showing 94% bidding 3NT. 1 vote for club transfer and 17 for 3NT.

Update: Poll is now 26-1 in favor of 3NT

[smerriman]

virgosrock, on 2017-December-11, 18:49, said:

smerriman does your sim include RHO ducking with Kx?

vrock

"based on double dummy analysis", so no.

But what wbartley says is true; if you force the defense to lead their longest/strongest suit, the probability of 3NT rises to about 60%, which is probably more than it'd go down based on the Kx duck.

[LBengtsson]

3nt 3nt 3nt 3nt 3nt 3nt 3nt 3nt 3nt 3nt 3nt 3nt 3nt 3nt 3nt 3nt 3nt 3♣?

i bid 3nt 110% of time