[rhm]

hotShot, on 2011-January-04, 07:53, said:

rhm showed that a hand that is better than OP's west hand (because it has 6+♥) will only have a 29.5% chance for game opposite likely east hands when opener has a precision 2♣ opening and his partner passes.
His results indicate that that such a west hand should bid over 2♣ but east should usually not bid on.

It is logical to conclude that east should pass, if west can have weaker hands than this.

The simulation does not say if the actual west hand should bid over 2♣.

My problem with this simulation is that the posting says nothing about the the east hand.



If the east hand is set to what it actually is on the OP, than the best move for East is to pass 2♥.

If the east hands are unrestricted in this simulation, than the results contain a lot of east hands that are weaker than the posted one.

The East hand in the simulation was set to what is the actual one on the OP.
Sorry if that was not clear.

Rainer Herrmann

[Fluffy]

rhm, on 2011-January-04, 08:29, said:

The East hand in the simulation was set to what is the actual one on the OP.
Sorry if that was not clear.

Rainer Herrmann

If you set one hand then double dummy simulations won't be equal as you though form the OK large examples, there are certain holdings (for example AKJ10 and AKJ9 combined) that favour declarer's play, while others don't help him much. East's actual hand is a good examplo of the non helping IMO, catching ♦Qx and some picky heart holdings but nothing else.

[awm]

I generated a bunch of hands and took a look single dummy. My results seem quite different from Rainer's results, indicating that game is very often good and that 3♥ rarely fails. I believe there are three main differences between my approach and his.

(1) I specified that overcaller has 11+ points and 6+♥. He seems to be allowing hands as weak as 9 hcp.
(2) I didn't set any requirement about responder's club length. Many of the really good game hands have opener holding singleton or small doubleton club. I don't think it's obvious that responder always raises with club length on this auction, especially since he seems marked with extremely light values.
(3) My single-dummy analysis might be different from double-dummy. There are very few hands in this set where our side has a real "play problem" whereas there were a number of hands where a non-intuitive diamond lead at trick one seems to be the only possible setting defense.

[MrAce]

awm, on 2011-January-04, 12:59, said:

I generated a bunch of hands and took a look single dummy. My results seem quite different from Rainer's results, indicating that game is very often good and that 3♥ rarely fails. I believe there are three main differences between my approach and his.

(1) I specified that overcaller has 11+ points and 6+♥. He seems to be allowing hands as weak as 9 hcp.
(2) I didn't set any requirement about responder's club length. Many of the really good game hands have opener holding singleton or small doubleton club. I don't think it's obvious that responder always raises with club length on this auction, especially since he seems marked with extremely light values.
(3) My single-dummy analysis might be different from double-dummy. There are very few hands in this set where our side has a real "play problem" whereas there were a number of hands where a non-intuitive diamond lead at trick one seems to be the only possible setting defense.

Nice, thats what i was wondering because u used more realistic restrictions. Rainer's test put too much meaning into responder's silence and i wrote b4 that it is very deceptive and not as reliable as he thinks it is. I also wrote that as did someone else, double dummy defense in 3♥ or 4♥ kinda contracts works much better for the defense side than for declarer.

Also, i believe u can add 9-10 hcp hands and still get the same result. Because someone who overcalls at 2 level with a 9 hcp, will not bid it on every 6 cards ♥ and every 9 hcps. More likely to hold 2 aces like ATx AJxxxx x xxx or similar 9 hcps. Not a Qxx AJxxxx x Qxx. This is very important and computer will not pay attention to the difference unless there is a way u tell it to do so.

[mfa1010]

Why is nobody talking about the opponents balancing anyway? It seems naive to base a lot on hoping to buy it in 2♥ making 8 when partner is minimum so that we have a 9-card fit and barely half the deck. Support with support. That will get us to game when partner has a good hand or when it really improves after being raised.

[whereagles]

I had a look at the dbl dummy site and I find the conclusion very interesting. It sure helps getting some arguments settled via simulations

[gwnn]

we need more simulations! who is right: awm or rhm? rhm or awm?

[gnasher]

gwnn, on 2011-January-05, 19:16, said:

we need more simulations! who is right: awm or rhm? rhm or awm?

Both of them, of course. The whole point of a simulation is to prove that you're right, so with two well-designed simulations we have two people who are right.

Edit:

This post has been edited by gnasher: 2011-January-06, 04:38

[gwnn]

If by "you" you mean me specifically, I agree. Therefore rhm's simulation missed the point.

[rhm]

gnasher, on 2011-January-06, 04:37, said:

Both of them, of course. The whole point of a simulation is to prove that you're right, so with two well-designed simulations we have two people who are right.

Edit:

I have not manipulated my assumptions in any way to prove my point, not least because I am not interested in winning the argument, but whether my judgment is right or not.
In the past I have done many simulations which showed I was either too optimistic or pessimistic and this is always enlightening. I think I have learned something from them.

However, my observation on this forum is, no matter what simulation you do, the ones whose judgment is in conflict with the results of the simulation will always find some silly arguments or use arguments long refuted to claim the results of the simulation is void or meaningless.

It is a fact of life that most people will rather doubt the facts than change their mind.

Rainer Herrmann

[655321]

awm, on 2011-January-04, 12:59, said:

I generated a bunch of hands and took a look single dummy. My results seem quite different from Rainer's results, indicating that game is very often good and that 3♥ rarely fails. I believe there are three main differences between my approach and his.

(1) I specified that overcaller has 11+ points and 6+♥. He seems to be allowing hands as weak as 9 hcp.
(2) I didn't set any requirement about responder's club length. Many of the really good game hands have opener holding singleton or small doubleton club. I don't think it's obvious that responder always raises with club length on this auction, especially since he seems marked with extremely light values.
(3) My single-dummy analysis might be different from double-dummy. There are very few hands in this set where our side has a real "play problem" whereas there were a number of hands where a non-intuitive diamond lead at trick one seems to be the only possible setting defense.

rhm, on 2011-January-07, 07:42, said:

However, my observation on this forum is, no matter what simulation you do, the ones whose judgment is in conflict with the results of the simulation will always find some silly arguments or use arguments long refuted to claim the results of the simulation is void or meaningless.

It is a fact of life that most people will rather doubt the facts than change their mind.

Rainer Herrmann

QED.

You admitted to MrAce in an earlier post that you didn't even check your results by hand to make sure that your parameters produced sensible hands. AWM ran a simulation, checked his results manually, and got very different answers. Your response was what we have come to expect from you, you ignored awm's post and instead implied that everyone on this forum who disagrees with your opinion (aka the facts!) is stupid.

My bridge judgment tells me that passing partner's vulnerable 2 level overcall on a decent fitting 10 count is terrible. Good players in this thread have said the same thing.

Obviously my bridge is not so wonderful that my judgment is infallible, yet if someone (especially someone who has a track record on bridge forums for bidding way outside mainstream expert practice) says, 'look, pass is best and I have run a simulation that proves it', my reaction is not to think that my bridge judgment in this common situation is completely wrong, my reaction is to assume that there were problems with the simulation. I don't see this as doubting the facts, I see this as trusting my years of experience at the bridge table ahead of a random internet poster.

[cherdano]

cherdano, on 2011-January-03, 05:15, said:

Why do you call a 3H raise an invite? I would call it a raise, promising less than invitational values, since with invitational values advancer would bid 3C.
But 655321 already told you that and you ignored it, so I am not sure there is much point in saying it again.

rhm, on 2011-January-03, 05:48, said:

I am not sure what you are aiming at. Are you claiming that a ♥ raise is preemptive and non-invitational or do you like splitting hairs?

The purpose of language is communication. If you use a word ("invitation") in a sense that does not coincide with the way the rest of the world uses it, then that makes communication difficult.

For me, for 655321 and for anyone else I know, a game invitational bid is a bid that invites partner to bid on with the majority of his hands. The raise to 3H is just a raise, partner needs extras to bid on. Since most hands are minimum, this means that most hands will pass the raise to 3H.

There are more obvious problems with your simulation, e.g. you don't really take RHO's pass into account, and your losses from raising to 3H don't take into account that some of the time LHO would have balanced. Also, maybe this particular hand plays better single-dummy than double-dummy. You apparently didn't even consider these pretty obvious possible flaws. Claiming your simulation results are "facts" that decide the question whether to raise make it even more obvious that you are blind to the possible problems of double-dummy simulations.

Hence, if I have to trust someone blindly, I will rather trust bridge judgment of experienced players instead of your simulations.

[rhm]

655321, on 2011-January-07, 09:08, said:

QED.

You admitted to MrAce in an earlier post that you didn't even check your results by hand to make sure that your parameters produced sensible hands. AWM ran a simulation, checked his results manually, and got very different answers. Your response was what we have come to expect from you, you ignored awm's post and instead implied that everyone on this forum who disagrees with your opinion (aka the facts!) is stupid.

My bridge judgment tells me that passing partner's vulnerable 2 level overcall on a decent fitting 10 count is terrible. Good players in this thread have said the same thing.

Obviously my bridge is not so wonderful that my judgment is infallible, yet if someone (especially someone who has a track record on bridge forums for bidding way outside mainstream expert practice) says, 'look, pass is best and I have run a simulation that proves it', my reaction is not to think that my bridge judgment in this common situation is completely wrong, my reaction is to assume that there were problems with the simulation. I don't see this as doubting the facts, I see this as trusting my years of experience at the bridge table ahead of a random internet poster.

Your answer shows that you do not know what a computer simulation is, neither do you seem to have a clue about statistics.
To do meaningful simulations, which can claim to be representative, means to use sample sizes to reduce the margin of error, which are that large, that you can not possibly check the results on all deals by hand, because the effort is just too high.
That's what computers are for. I said I generated 1000 deals to keep the margin of error small. If I would have checked carefully every single deal by hand I would still be looking at them
However, on every simulation I check a few deals whether I would in deed have over-called 2♥ or open 2♣ to prove that my assumptions are reasonable.
From many runs I know by now that I can trust the randomness of my (commercial) software and the double dummy analysis of deep finesse.

AWM claimed to have done a simulation. Different to me he did not specify his assumptions, he did not say how many deals he generated and looked at. This is very dubious to start with.
Since he claimed he looked at them by hand, hopefully carefully, it can not have been many for the reasons above. This method is far more subjective and biased than large sample double dummy analysis by software.
Even if his analysis by hand is correct, which is difficult to accomplish, and the few deals have been generated randomly according to the specifications, the margin of error will still be high.

Rainer Herrmann

[hotShot]

655321, on 2011-January-07, 09:08, said:

QED.

You admitted to MrAce in an earlier post that you didn't even check your results by hand to make sure that your parameters produced sensible hands. AWM ran a simulation, checked his results manually, and got very different answers. Your response was what we have come to expect from you, you ignored awm's post and instead implied that everyone on this forum who disagrees with your opinion (aka the facts!) is stupid.

My bridge judgment tells me that passing partner's vulnerable 2 level overcall on a decent fitting 10 count is terrible. Good players in this thread have said the same thing.

Obviously my bridge is not so wonderful that my judgment is infallible, yet if someone (especially someone who has a track record on bridge forums for bidding way outside mainstream expert practice) says, 'look, pass is best and I have run a simulation that proves it', my reaction is not to think that my bridge judgment in this common situation is completely wrong, my reaction is to assume that there were problems with the simulation. I don't see this as doubting the facts, I see this as trusting my years of experience at the bridge table ahead of a random internet poster.

I think you are not fair. Both rhm and awm have specified (most of) their simulation parameters, Rainer offered to rerun the simulation if someone had objections to the ones he made, but nobody posted usable parameters.

Rainer specified overcaller with "No 5 card ♠ suit, either 6 cards in ♥ and 10 to 17 HCP or 7 cards in ♥ and 10 to 15 HCP".
Adam specified overcaller with "11+ points and 6+♥."

Adam's setting should produce more games than Rainer's.

Rainer specified that North should not have less than 4♣, while Adam did not restrict responders ♣ length.
This is problematic, because allowing responder to have more ♣ will shorten the average ♣ length of West shifting the result slightly in favor of EW's ♥ contract.
But his effect should not be bigger than 0.5 percent points.

Rainer used "1000 random deals, double dummy results with West declarer in a ♥ contract", while Adam used: "My single-dummy analysis might be different from double-dummy. There are very few hands in this set where our side has a real "play problem" whereas there were a number of hands where a non-intuitive diamond lead at trick one seems to be the only possible setting defense."

This is the interesting part, from all the critique made about Rainer's simulation the only relevant thing is that North's ♦ lead is counter intuitive.
Obviously ♦AQ (T98) are more likely to be in the South hand than in North hand and the double dummy solver will always find the ♦ lead.
So it is possible that in this specific setting the double dummy solver could deviate more from the human play than the average 0.1 tricks.

Rainer stated that about 62% of the time you can make 3♥, Adams little extra strength and a non intuitive lead could easily shift a lot of these to making 4.

[OleBerg]

Just to split another semantic hair:

It seems it could be usefull to distinguish betweem raises that are consultative and raises that are unilateral.

It seems like many here would both call:

(2♣) - 2♥ - (Pass) - 3♥

and

2♥ - (Pass) - 3♥

a "single raise".

But there is quite some difference;

In the first sequence, partner is expected to go to game on certain hands, and is allowed to expect a reasonably well-defined hand-type.

In the second sequence, opener is (in principle) forbidden to go to game, and cannot expect partnes hand to be very well-defined, neither concerning strength nor distribution; it may be very weak, or it may be just below a game-invite.

[jamegumb]

I'd add overcalling hands with 12-18 HCP, 5♥, and 3+♣ to the simulations.

[gwnn]

2♥-p-3♥

is not a single raise, it's a signoff.

[gnasher]

rhm, on 2011-January-07, 07:42, said:

I have not manipulated my assumptions in any way to prove my point, not least because I am not interested in winning the argument, but whether my judgment is right or not.

I'm sure you haven't. When I edited my post to add a smiley, I did so in order to indicate that my post was a joke. Sorry if that wasn't clear.

[rhm]

gwnn, on 2011-January-07, 11:21, said:

2♥-p-3♥

is not a single raise, it's a signoff.

What about speaking English for a change and calling both 3♥ bids raises, but one of these raises is a sign-off while the other one is invitational to 4♥?

Rainer Herrmann

[MrAce]