[relknes]

 akhare, on 2013-January-28, 17:19, said:

The following redeal script indicated that 5C is a winner in a 10000 hand simulation.

What percentage of the time did each contract make? Was it close, or was it a blowout?

[akhare]

 relknes, on 2013-January-28, 18:04, said:

What percentage of the time did each contract make? Was it close, or was it a blowout?

Actually, I think I stand corrected. My personal bias towards 5♣ likely caused to misinterpret the data in this table and as see it now, 4♠ > 3N > 5♣.

BTW, the formatting doesn't quite come out right, but it's essentially a payoff of one contract vs. another. So, below 3N gains +0.72 over 5C and loses by -0.51 to 4S etc. with an implied +0 against itself.

bid3N.....bid5C.....bid4S
bid3N.....+0.72.....-0.51
..........(0.07).....(0.08)
bid5C...-0.72........-1.02
........(0.07).......(0.07)
bid4S....+0.51.......+1.02
........(0.08).......(0.07)

Richard brought up an excellent point regarding the statistical significance of the data. A better simulation will assign the shape (4=5=1=3) and RPs / HCPs to the North hand without the specific hand in question.

I will try and modify the script to do it, but am not very comfortable with the redeal Python syntax -- perhaps, antonylee can help?

[antonylee]

Yes, I was wondering why I had 3N clearly higher than 5♣ in my own sims...
Here is the IMPs payoff table (10,000 hands), also including 4♥. All contracts are assumed to be played by South (I don't think it matter much here). Also you need to pass True as second argument to dd_score (for vulnerability).

	bid3N	bid4H	bid4S	bid5C	
bid3N		-0.66	-0.39	+0.77
		(0.03)	(0.03)	(0.02)
bid4H	+0.66		+0.30	+1.24
	(0.03)		(0.03)	(0.02)
bid4S	+0.39	-0.30		+0.93
	(0.03)	(0.03)		(0.03)
bid5C	-0.77	-1.24	-0.93	
	(0.02)	(0.02)	(0.03)

(hint: use [ code ] for the formatting)
So it's basically 4♥ > 4♠ > 3N > 5♣ (and numbers in parentheses are standard deviations, so to first approximation all differences are statistically significant).

On the other hand I think DD sims should be taken with a grain of salt here because indeed 3N becomes much better as soon as you guess the clubs right. I wouldn't be surprised if it fared worse than 5♣ "in practice".

As for syntax I basically ran your simulation with an extra line for 4♥. I would be happy to implement a simpler syntax but I'll need your suggestions as this one is the best (that caters for everything I want to sim) I've found so far (granted, I did not put much thought in it either).

[awm]

There are a number of choices for us in the auction, I will go with this one:

IMprecision:

1♣ - 1♥ (strong, 4+♠ with 5+ hcp and 2-6RP)
1♠ - 2♥ (various hands including 3+♠, 4♠ and 5+♥)
2NT - 3♣ (NF misfitty, natural patterning out and GF)
3♦ - 3NT (nothing wasted in diamonds, no extra shape/values and potentially NF)
4♥ - Pass (must be Hx, reasonable heart suit)

Some decisions:

Opener could rebid 1NT over 1♥ as this hand will not play spectacularly in a 4-3 fit and has some issues when responder shows hearts. Then the auction would proceed 2♣(stayman)-2♦-2♥(puppet)-2♠-3♣(both majors short diamonds)-3♦(no wastage)-3♥(longer major)-3♠(three spades) and we would scramble to one of 5♣/4♥/4♠ (I think 4♥ most likely).

Opener could continue relaying over 2♥ and hope for the best since he does have "working" values; then responder would show pattern via 3♦ and opener would have to guess which game to play (generally we don't play 3NT here, but any of 5♣/4♥/4♠ seem possible).

On our given auction, opener could pass 3NT or bid 4♣ or 4♠. However I think 4♥ (showing Hx) is best because it leaves responder the option to remove it when holding weak hearts.

[rbforster]

 akhare, on 2013-January-28, 17:19, said:

The following redeal script...

predeal = {"S": H("AK4 Q5 A85 AJT83"),
"N": H("JT32 AJ732 3 K65")}

Akhare - I think you're assuming the actual North hand, not sure if that matters. The situation faced by most precision relayers was that South knew North had 4513 min GF, say 8-10 hcp roughly. It would be interesting to sim across the various North hands subject to those shape/value constraints to see what game contracts seemed best.

[Kungsgeten]

 straube, on 2013-January-28, 01:30, said:

You are South and dealer. V vs V. Imps. Uncontested auction.

...........................JT32
...........................AJ732
...........................3
...........................K65
?.....................................................?
?.....................................................?
?.....................................................?
?.....................................................?
...........................AK4
...........................Q5
...........................A85
...........................AJT83

Pass 2000
pass - 1D; (0-7 / 17+ ; 4+ hearts, 8-16, may have longer minor)
1H - 1S; (relay, may be weak ; natural, may be 44(14))
1N - 2D; (GF relay ; short diamonds)
2H - 2S; (R ; 4-5-1-3)
2N - 3D; (ace ask ; 1 ace)
3H - 3N; (king ask ; 1 king)
pass

[Zelandakh]

 relknes, on 2013-January-28, 10:28, said:

edit: after looking at the hands, I wonder if north should have tried a game in hearts on the 5-2 fit.

Why would you do this without looking at the hands?

 Kungsgeten, on 2013-January-29, 01:41, said:

Pass 2000

How would you bid it if North has ♠QJTx/ ♥AJTxx/ ♦x/ ♣KQx?

[antonylee]

 rbforster, on 2013-January-28, 22:02, said:

Akhare - I think you're assuming the actual North hand, not sure if that matters. The situation faced by most precision relayers was that South knew North had 4513 min GF, say 8-10 hcp roughly. It would be interesting to sim across the various North hands subject to those shape/value constraints to see what game contracts seemed best.

Here we go for 10,000 deals with North constrained to 4513 and 8-10 HCP. I only list the probability that a contract makes (double dummy):
5♣ 90%
4♥ 88%
4♠ 87%
3N 80%
So except for 3N all other games are basically equally good (or rather, the 2-3% difference is probably negligible compared to the approximation of double-dummy play); and 5♣ goes up again.

[Zelandakh]

Could you do the same for 9-11 too please Antony?

[qplus10]

1♣1♥(5+♥gf)
3nt

[Free]

 qplus10, on 2013-January-29, 08:23, said:

1♣1♥(5+♥gf)
3nt

Proof that this system really sucks or that we don't get a full explanation

[relknes]

 Zelandakh, on 2013-January-29, 03:15, said:

Why would you do this without looking at the hands?

I wouldn't. I had a momentary brain lapse after looking at the hands.
I guess the better question for me to ask is, would it be worth giving up one of my slam tools in this auction to gain the ability to explore a major suit game on a known 5-2 fit.

[Zelandakh]

 relknes, on 2013-January-29, 09:41, said:

I wouldn't. I had a momentary brain lapse after looking at the hands.
I guess the better question for me to ask is, would it be worth giving up one of my slam tools in this auction to gain the ability to explore a major suit game on a known 5-2 fit.

Yeah, this is easy to do. Sometimes I wish I could reconsider some of my table bids after I see the hands too!

As for a solution, well how do you play:

1♣ - 1♥;
1NT - 2♠?

If this is natural then you can continue:

2NT - 3♣
to pattern out. Now you have reached a similar position to the relayers. Basically, the simplest solution is to have some way for North to show all 3 suits (and hence shortage in the fourth) and then leave the decision to South.

[straube]

 antonylee, on 2013-January-29, 03:49, said:

Here we go for 10,000 deals with North constrained to 4513 and 8-10 HCP. I only list the probability that a contract makes (double dummy):
5♣ 90%
4♥ 88%
4♠ 87%
3N 80%
So except for 3N all other games are basically equally good (or rather, the 2-3% difference is probably negligible compared to the approximation of double-dummy play); and 5♣ goes up again.

Thanks for running this. I found this really useful.

[relknes]

 Zelandakh, on 2013-January-29, 09:54, said:

Yeah, this is easy to do. Sometimes I wish I could reconsider some of my table bids after I see the hands too!

As for a solution, well how do you play:

1♣ - 1♥;
1NT - 2♠?

If this is natural then you can continue:

2NT - 3♣
to pattern out. Now you have reached a similar position to the relayers. Basically, the simplest solution is to have some way for North to show all 3 suits (and hence shortage in the fourth) and then leave the decision to South.

Currently, 1♣ - 1♥ - 1NT - 2♠ would be minor suit stayman.
We play simmilar conventions over 1♣-1x-1N as we do over a 1N opener, except that responder rebidding their suit after giving a positive response sets trump absolutely, opener should show 3 card support for responder's major if responder bids stayman (new minor forcing), and 3m shows a 5-5 hand instead of invitational with a strong minor. If we wanted to show 5 hearts and 4 spades, we could bid 1♣ - 1♥ - 1NT - 2♦ (transfer) - 2♥ - 2♠ after which we could continue on as you suggested.
Thanks for the imput. Looks like we had the tools after all...
I have added to my original post, but left the previous auctions intact so people understand what we are talking about.

[jack502]

1♣2♥
3♣3n
4♥

2♥semi positive 5+ hearts 4 spades
3♣nat 15-18

[antonylee]

 antonylee, on 2013-January-29, 03:49, said:

Here we go for 10,000 deals with North constrained to 4513 and 8-10 HCP. I only list the probability that a contract makes (double dummy):
5♣ 90%
4♥ 88%
4♠ 87%
3N 80%
So except for 3N all other games are basically equally good (or rather, the 2-3% difference is probably negligible compared to the approximation of double-dummy play); and 5♣ goes up again.

And for 9-11 (and 4513), as requested by Zelandakh
5♣ 95%
4♥ 93%
4♠ 93%
3N 88%
Again 3N is a bit behind (less, though) but the three other games are basically equally good.

[Mbodell]

5♣S for tosr

1♣-1♥ : 16+ bal, 18+ unbal; 8+hcp, 2+ctls, 4+♠
1♠-1nt : relay; 4+♥
2♣-2♥ : relay; 4♠5+♥
2♠-2nt : relay; short ♦
3♣-3♦ : relay; 4=5=1=3
3♥-3nt : relay; 3 ctls
4♦-4♥ : ->4♥ to sign off somewhere; obeying
5♣ : to play

Partner did think about trying a major but figured clubs was likely to be better. All you know is 8+ points (but probably not 13 or someting) 4=5=1=3 and for controls either the missing 3K or ♥A and some K (and at least a J if it is A+K).

[Zelandakh]

 antonylee, on 2013-January-29, 13:34, said:

And for 9-11 (and 4513), as requested by Zelandakh

Thanks very much Antony!

From the results, it seems like we should probably be playing 5♣ at IMPs and 4♥ at MPs.

[Siegmund]

Unassuming has a boring auction: 1C-1H-1NT (15-18 balanced), checkback stayman method of your choice, 3NT.

Malfoir backs itself into a corner but lands on its feet:

1C-1S (if opener has the 11-18 hand with 4 spades and <4 hearts, spades are going to be trump, so might as well conceal them, and take control of the auction)
1NT-2D (15-18 no 4CM / inv+ transfer, can be 4S5H or 5S5H)
3C-4S (maximum, 3 spades but usually not 3 hearts, values in clubs that might be wasted opposite shortness)
and now responder really ought to be scared of playing 3NT, if opener chose to advertise club values rather than diamond values.