[mikeh]

 whereagles, on 2014-September-21, 11:56, said:

I know of two common agreements when it comes to doubles of splinters:

1. Asks for a lead of the suit BELOW the singleton.
2. Shows length in the splinter suit, suggests a sac.

.

those are common agreements when partner has been silent. We are, of course, speaking here of a completely different situation.

I stand by my original and, as it happens, 'works-at-the-table', decision to pass, but I have been persuaded that the optimal agreement here is to double to suggest a save. I agree that such usage stands out over all over options and had I understood that double would have been taken as such I would double.

As it happened, I (and I think most of us) had seen the problem as offering a choice between the save and the pass, and bridge players being aggressive generally, certainly when posting on a forum, the majority, and the players at the table, opted for the aggressive choice. I suspect that virtually all experts would agree that neither 4♠ nor pass are attractive, and that were a middle ground available, that allowed us to pass the decision to partner, such would have been a clear choice.

So: I still pass until I discuss this meaning with partner, after which I double, feeling happy.

[lamford]

 nige1, on 2014-September-20, 17:20, said:

who are the strong players that agreed with you, Paul?

Er, I don't think I claimed that anyone particularly agreed with me, as I had the benefit of seeing all four hands. I am persuaded by PhilKing's argument that double invites a save, and partner will pass out 4H and try to beat it. If he leads a spade he will fail; if he leads a club he should still fail, but two top declarers went off.

Deep Finesse certainly agrees with me in that 4Sx is -800. Declarer had a trump suit of Kxxxx opposite J9xx and did not have the ten, so will probably make 4H. I think 4S is wrong, and now think I was wrong to agree with Pass. The consultative double looks like a standout.

[whereagles]

 mikeh, on 2014-September-21, 12:18, said:

So: I still pass until I discuss this meaning with partner, after which I double, feeling happy.

That is usually a good idea, yes.

[nige1]

 lamford, on 2014-September-21, 13:52, said:

Er, I don't think I claimed that anyone particularly agreed with me, as I had the benefit of seeing all four hands. I am persuaded by PhilKing's argument that double invites a save, and partner will pass out 4H and try to beat it. If he leads a spade he will fail; if he leads a club he should still fail, but two top declarers went off. Deep Finesse certainly agrees with me in that 4Sx is -800. Declarer had a trump suit of Kxxxx opposite J9xx and did not have the ten, so will probably make 4H. I think 4S is wrong, and now think I was wrong to agree with Pass. The consultative double looks like a standout.

Why "sackcloth and ashes" if you've nothing for which to be contrite? Did the "strong players" make a successful choice?

[MrAce]

I still can't believe the pass to be correct assuming that we do not have the option where DBL shows what Phil said. When we take out the double as an option, 4♠ and pass are only options and I believe that 4♠ stands out and it's not even close imo.
Yes, I am aware that pd is holding 1-2 hearts. So let's give him 2 to make it for the favour of passers. And yes I also know pd holds about 7-8 hcp if someone is not overbidding massively. That marks pd with only 5 ♠, with 6 he would probably start a WJO. What about his minor holdings? Let's also give pd the worst case scenario and give him 3-3 minors, a totally balanced hand. I am aware that under these circumstances we will go -800. And I am guessing that's what happened. -800 vs -620 and yes there are rare times in this auction that they may go down.

But how about if you change the minor holdings.make them 4-2 or 5-1. Or make opponents hold a 10 card fit instead of only 9. Or think about the hands where opponents (not knowing that we both have totally balanced hands) took another shot at 5 level vulnerable game, vs a 4♠ nun vuln save. You never see this happen? 4♥ maybe cold and 5 may be too high. If we pass now, pd will pass with 5233 but also will pass with 5134 5143 5125 5215 5251....
Of course, if agreed, DBL is the best of both worlds when it gives an option to 1♠ bidder. He will pass with 5233 hands, obviously, as Lamford said and he will know to take action when holding 5134 5215 etc.

Perhaps a simulation can show us what's best (other than dbl) pass or 4♠. Give pd exactly 5 ♠, 0-1-2 ♥ 6-8 hcp. LHO 12-14 5-6-7 hearts, RHO 4-5-6 hearts + stiff or void spades, 9-12 hcp.

[mikeh]

 MrAce, on 2014-September-21, 18:47, said:

I still can't believe the pass to be correct assuming that we do not have the option where DBL shows what Phil said. When we take out the double as an option, 4♠ and pass are only options and I believe that 4♠ stands out and it's not even close imo.
Yes, I am aware that pd is holding 1-2 hearts. So let's give him 2 to make it for the favour of passers. And yes I also know pd holds about 7-8 hcp if someone is not overbidding massively. That marks pd with only 5 ♠, with 6 he would probably start a WJO. What about his minor holdings? Let's also give pd the worst case scenario and give him 3-3 minors, a totally balanced hand. I am aware that under these circumstances we will go -800. And I am guessing that's what happened. -800 vs -620 and yes there are rare times in this auction that they may go down.

But how about if you change the minor holdings.make them 4-2 or 5-1. Or make opponents hold a 10 card fit instead of only 9. Or think about the hands where opponents (not knowing that we both have totally balanced hands) took another shot at 5 level vulnerable game, vs a 4♠ nun vuln save. You never see this happen? 4♥ maybe cold and 5 may be too high. If we pass now, pd will pass with 5233 but also will pass with 5134 5143 5125 5215 5251....
Of course, if agreed, DBL is the best of both worlds when it gives an option to 1♠ bidder. He will pass with 5233 hands, obviously, as Lamford said and he will know to take action when holding 5134 5215 etc.

Perhaps a simulation can show us what's best (other than dbl) pass or 4♠. Give pd exactly 5 ♠, 0-1-2 ♥ 6-8 hcp. LHO 12-14 5-6-7 hearts, RHO 4-5-6 hearts + stiff or void spades, 9-12 hcp.

I really don't like your constraints. All kinds of posters use that silly rule of 20, but even assuming that the presumed expert opps aren't that bad, it is still common to open shapely 10 counts, and almost everyone opens decent 11's with a major. Then most splinter aggressively, so while I don't 'expect' 10 hcp from partner, I won't the director on the opps if they have a combined 21-22 hcp, and nor would you, or any other good player. This means that partner will have between 7-11 hcp...fwiw I don't think non-space consuming overcalls on AQxxx and out is winning bridge even at favourable.

So give partner 7-11 and run your sims. Otherwise you are stacking the deck against going plus on defence, which is unfair.

[wynsten]

Looks to me like this is could be a 17 trick hand (each side having nine trumps, with spades 3-1, and hearts 2-2). It's the 2-2 hearts that denies NS a ruff and therefore plays a trick less than expected.

Suppose N has AQxxx-Kx-xxx-xxx and that club Ace lies over South's King. Three tricks on defense; seven as declarer. So 4S could lead to a bad result (-800 vs -620)

Give NS another trick, say AQxxx-Kx-xxx-Qxx. Now 4S is only down 2, but 4H is also down. 4S is a phantom sac.

It seems to me that 4S is likely to be a losing bid. As for choosing between P and Dbl, I lean to giving less information to opponents here than more to partner, so I pass.

Oops - when I posted the above, I thought NS were vulnerable and off 3 was -800. Non-vulnerable 4S has a chance to be right (-500 vs -620) or wrong (-800 vs -650). I probably still pass, but it is not so clear.

[lamford]

 MrAce, on 2014-September-21, 18:47, said:

Perhaps a simulation can show us what's best (other than dbl) pass or 4♠. Give pd exactly 5 ♠, 0-1-2 ♥ 6-8 hcp. LHO 12-14 5-6-7 hearts, RHO 4-5-6 hearts + stiff or void spades, 9-12 hcp.

I will try to do one later in the week, but too busy at the moment.

[ArtK78]

I believe that a lot of posters are using the wrong approach in sacrifice situations.

Look at this hand. A number of comments seem to imply that if you go for -500 and the opps make their game, you have won some great victory. That may be true at matchpoints, assuming that a great percentage of the field is in game on the opps cards. But at IMPs, -500 vs. -620 is only 3 IMPs - not the greatest of victories. And you need quite a parlay to get this result:

(1) You must be able to take 7 tricks in 4♠;
(2) Your opps must be making their contract;
(3) 4♠x will be the final contract.

You may also win a lot by bidding 4♠ when the opps bid on and go down or if you go down less than 3 tricks.

How can bidding 4♠ lose? The obvious way is that you go down more than 3 tricks.

Another way that bidding 4♠ can lose is if the opps were destined to go down in 4♥. Given your hand, that is not impossible.

Looking at your hand, I would guess that the opps are unlikely to bid on; however, if you go down less than 3 tricks on this hand it is probably more likelty that the opps can bid on and make, so sacrificing at 4♠ may offer the opponents a fielder's choice - they get to choose from multiple winning options.

On this hand, I would guess that the chances of going plus in 4♥ are good enough to defend, and that if you do sactifice you probably will not gain much assuming that you come out ahead.

Yesterday, I played in a small sectional swiss team. One of our opponents was a team that liked to bid a lot. On two of the seven hands in our match, they took favorability vulnerability sacs against our games. One was a push (the sacrifice was a normal action). On the other they lost 12 IMPs, as their sac went for 500 and our game was down one.

[FrancesHinden]

 MrAce, on 2014-September-21, 18:47, said:

Perhaps a simulation can show us what's best (other than dbl) pass or 4♠. Give pd exactly 5 ♠, 0-1-2 ♥ 6-8 hcp. LHO 12-14 5-6-7 hearts, RHO 4-5-6 hearts + stiff or void spades, 9-12 hcp.

6-8 HCP?? I also played this hand, and I'll tell you for free that partner has an 11-count. Opener also has 11. Responder has - wait for it - 10.
We defended 4H, on a different auction, and my spade lead made it easy. Teammates took 800.

[MrAce]

 FrancesHinden, on 2014-September-22, 15:09, said:

6-8 HCP?? I also played this hand, and I'll tell you for free that partner has an 11-count. Opener also has 11. Responder has - wait for it - 10.
We defended 4H, on a different auction, and my spade lead made it easy. Teammates took 800.

We are not discussing what is the right action this time in particular. We are debating which is the right action in this position. Without your free tell, I already guessed 4♠ was not gonna be the winner on this particular hand.(Read my first reply) We are not even debating whether 4♠ is better than DBL or not, we are comparing 4♠ and pass if we are forced to choose one. I already like the idea and usage of DBL mentioned by Andy and Phil. No argument there on my end.

If you guys are telling me that 22 hcp is the most expected strength of opponents, when they are bidding game freely and suggesting slam, and we should not bid 4♠ over 4♥ with 9 spades when they are known to have 9+ hearts, white vs red, more power to you.

The way Mike suggests (opener can have 10 hcp) is not fair either. When we tell this to simulator, it will count all the 10 hcps regardless. But people do not open 10 hcp hands without a reason. They usually hold 5-5 6-4 ar at least 5431 type of hands. Mike also says "I do not like AQxxx and out overcalls" This is a ruff definition, and I don't like it either if it is 5332. But I sure like it w/r especially when I have unbalanced hand. I would definitely overcall AQxxx x xx JTxxx, who would not? How could I know the guy who splintered, upgraded his hand due to singleton which he is showing anyway. Are we supposed to just pass and let them play 4♥, just incase this hand be one of those where they can not make and we can take 800? Just incase pd may have a balanced hand? Just incase they have only 9 card fit? Just incase they always know when to bid 5 ♥ and when not to? At these colors? Sorry it doesn't feel right to me.

[mikeh]

 MrAce, on 2014-September-22, 20:13, said:

We are not discussing what is the right action this time in particular. We are debating which is the right action in this position. Without your free tell, I already guessed 4♠ was not gonna be the winner on this particular hand.(Read my first reply) We are not even debating whether 4♠ is better than DBL or not, we are comparing 4♠ and pass if we are forced to choose one. I already like the idea and usage of DBL mentioned by Andy and Phil. No argument there on my end.

If you guys are telling me that 22 hcp is the most expected strength of opponents, when they are bidding game freely and suggesting slam, and we should not bid 4♠ over 4♥ with 9 spades when they are known to have 9+ hearts, white vs red, more power to you.

The way Mike suggests (opener can have 10 hcp) is not fair either. When we tell this to simulator, it will count all the 10 hcps regardless. But people do not open 10 hcp hands without a reason. They usually hold 5-5 6-4 ar at least 5431 type of hands. Mike also says "I do not like AQxxx and out overcalls" This is a ruff definition, and I don't like it either if it is 5332. But I sure like it w/r especially when I have unbalanced hand. I would definitely overcall AQxxx x xx JTxxx, who would not? How could I know the guy who splintered, upgraded his hand due to singleton which he is showing anyway. Are we supposed to just pass and let them play 4♥, just incase this hand be one of those where they can not make and we can take 800? Just incase pd may have a balanced hand? Just incase they have only 9 card fit? Just incase they always know when to bid 5 ♥ and when not to? At these colors? Sorry it doesn't feel right to me.

Timo

I think you missed the point. You stated that your idea of a simulation was to give partner 6-8 points. I and others, including the players who held the hands, think that is too low and too narrow a range. When I suggest opener might have 10 or 11, I am not suggesting that we run a simulation that has opener opening all hands with 10-11.

This issue is one reason why simulations are of very limited use. The reality is that no program of which I am aware seems capable of allowing really good simulation of borderline hands. That is why I often run fairly short simulations, with an eyeball look at the hands to weed out ones that don't measure up. That of course makes the simulation somewhat subjective which some would say is counter to the notion of simulations, but we can't have it both ways. It is definitely wrong to reject 9-11 counts for partner just because most 10-11 hands held by the opps won't (usually) be bid as the OP says they were. That doesn't mean that the opps don't hold weak hands that WOULD bid that way.

Finally, since when did anyone say that the opps can't have more than 22 hcp? I, for one, merely pointed out that the opps MIGHT have as few as 21-22. The fact that I appear to have read the situation correctly isn't proof of anything beyond that 'MIGHT'. 'MIGHT' isn't "DEFINITELY'

[MrAce]

Fair enough, I think you are right that I took it differently than your last post. When I assigned 6-8 hcp in this auction, I did not mean he can not have more. I believe rightly or wrongly that 7 or 8 hcp will be the most frequent hands he will hold.
Once again, it's not absurd to think my opponents will hold at least 23-24 hcp when they make a slam try, much more frequently than the possibility of holding 21-22 hcp.
You are probably right about the simulations to be deceptive, I will take your word for it since I do not have a simulator, and never used one. If you remember I had an older version of DM and I pm ed you because it wasn't working , which later the guy who sells it sent me a link to fix it but that didn't work either. Basically I have no experience with that tool whatsoever.

[lamford]

I did a simulation with the following parameters:
a) Play including the opening lead is 'perfect' for both sides
b) You are always doubled in Four Spades, and they never bid Five Hearts
c) The opener has five+ hearts and 11+ points
d) The overcaller has five or six spades and 7+ points
e) The splinterer (that sounds a horrible agent noun) has 7-11 with 0-1 spades and 4-5 hearts
f) You never double Four Hearts

The matrix was as follows:
Contract6 7 8 9 10 11 12 13
Hearts 0 11 103 343 352 168 23 0
Spades 55 218 353 276 93 5 0 0

What this jumble tells us is that given the parameters above 4H will only make 54.3% of the time. 4S will only make 9.8% of the time, but I don't think many were bidding it to make. The expected value of 4S is -228.18 and the expected value of defending 4H is -218.18. Remarkably close, and, as they say in some events, a difference of 10 is a draw!

However, included in the above are hands where overcaller has six spades. If he has only five then Four Spades obviously fares much worse, as we know trumps are 4-0 or 3-1. It seems to me that double will get partner to sacrifice, lawfully, whenever he has six spades, and view to defend unless he has a high ODR when he has five spades. The hands to avoid are the ones where we phantom for -300 or worse, and we do not gain enough for going for -500 against 620. The "flexible" double seems to be the winner here.

And can a moderator tell me how to tab the table above, please!

[lamford]

 FrancesHinden, on 2014-September-22, 15:09, said:

Teammates took 800.

That must have been one of your teammates' better results ...

[the_clown]

I would certainly double here partner can look at his hand and decide if sacrifice is a good idea. I cant see how double is not a spade raise, its way too important to give up on these auctions.

[wynsten]

 lamford, on 2014-September-23, 04:57, said:

I did a simulation with the following parameters:
a) Play including the opening lead is 'perfect' for both sides
b) You are always doubled in Four Spades, and they never bid Five Hearts
c) The opener has five+ hearts and 11+ points
d) The overcaller has five or six spades and 7+ points
e) The splinterer (that sounds a horrible agent noun) has 7-11 with 0-1 spades and 4-5 hearts
f) You never double Four Hearts

The matrix was as follows:
Contract6 7 8 9 10 11 12 13
Hearts 0 11 103 343 352 168 23 0
Spades 55 218 353 276 93 5 0 0

What this jumble tells us is that given the parameters above 4H will only make 54.3% of the time. 4S will only make 9.8% of the time, but I don't think many were bidding it to make. The expected value of 4S is -228.18 and the expected value of defending 4H is -218.18. Remarkably close, and, as they say in some events, a difference of 10 is a draw!

However, included in the above are hands where overcaller has six spades. If he has only five then Four Spades obviously fares much worse, as we know trumps are 4-0 or 3-1. It seems to me that double will get partner to sacrifice, lawfully, whenever he has six spades, and view to defend unless he has a high ODR when he has five spades. The hands to avoid are the ones where we phantom for -300 or worse, and we do not gain enough for going for -500 against 620. The "flexible" double seems to be the winner here.

So, after all that calculation, and assuming the accuracy of the various constraints, what is the expected IMP differential if 4S is bid at one table, and Pass at the other?
The downside of "Double" is that it communicates as much information to declarer as it does to partner, perhaps allowing him to hold his trumps to one loser missing AQxx. How does your simulation allow you to account for this?

[MrAce]

 lamford, on 2014-September-23, 04:57, said:

I did a simulation with the following parameters:
a) Play including the opening lead is 'perfect' for both sides
b) You are always doubled in Four Spades, and they never bid Five Hearts
c) The opener has five+ hearts and 11+ points
d) The overcaller has five or six spades and 7+ points
e) The splinterer (that sounds a horrible agent noun) has 7-11 with 0-1 spades and 4-5 hearts
f) You never double Four Hearts

The matrix was as follows:
Contract6 7 8 9 10 11 12 13
Hearts 0 11 103 343 352 168 23 0
Spades 55 218 353 276 93 5 0 0

What this jumble tells us is that given the parameters above 4H will only make 54.3% of the time. 4S will only make 9.8% of the time, but I don't think many were bidding it to make. The expected value of 4S is -228.18 and the expected value of defending 4H is -218.18. Remarkably close, and, as they say in some events, a difference of 10 is a draw!

However, included in the above are hands where overcaller has six spades. If he has only five then Four Spades obviously fares much worse, as we know trumps are 4-0 or 3-1. It seems to me that double will get partner to sacrifice, lawfully, whenever he has six spades, and view to defend unless he has a high ODR when he has five spades. The hands to avoid are the ones where we phantom for -300 or worse, and we do not gain enough for going for -500 against 620. The "flexible" double seems to be the winner here.

And can a moderator tell me how to tab the table above, please!

Thanks for the simulation. I think you were fair in general.

[lamford]

 wynsten, on 2014-September-23, 07:48, said:

So, after all that calculation, and assuming the accuracy of the various constraints, what is the expected IMP differential if 4S is bid at one table, and Pass at the other?
The downside of "Double" is that it communicates as much information to declarer as it does to partner, perhaps allowing him to hold his trumps to one loser missing AQxx. How does your simulation allow you to account for this?

If the choice were just between Pass and 4S, then there would be nothing to choose. However, double gains when partner judges well over 4H, and the double does help him to do this. It could be argued double helps declarer more than Pass does, in that it describes our hand quite well.

I don't think double says much about the trump holding, and the simulation does indeed only play perfectly as stated. Double says, "I don't want to bid 4S and I don't want to Pass".

"Just right", said Goldilocks.

[wynsten]

 lamford, on 2014-September-23, 09:58, said:

If the choice were just between Pass and 4S, then there would be nothing to choose.

OK - but I thought the point of the simulation was to determine this apparently self-evident fact. I just meant to point out that since the IMP function is non-linear, the expectation of the IMP (what we really need to know) does not necessarily equal the IMP of the expectation (which is what you were implying when you said the IMP of a ten point expected difference is zero).

As for what the double shows, I don't really know (despite the forum title, I'm no expert) so I took the extreme case that it showed everything. Hence the assistance in handling trumps.