Help on a simulation

[Fluffy]

Recently I've seen some members of the forum saying that Jxx in partner's 4 card suit is a good holding, I think exactly the contrary, so I would like to try a simualtion.

 

I haven't ever made a simulation, is it possible to make it compute a hand double dummy, and then compute the same hand A: removing the Jack for a small spot, B: removing the jack for a small spot but putting something worth 1 HCP on another suit.

 

If its possible please tell em with wich programs.

[P_Marlowe]

Hi,

 

it is.

 

You can use

 

http://bridge.thomasoandrews.com/deal/

 

I did not visit the side for a couple of months and voila, there is a new

feature, which is relevant for you, the programm has now a build in

DD solver, previously you needed the old GIB DD solver, which you

needed to track down on the net.

 

Now just deal the hand make your DD analysis, and modify the hand,

make another DD analysis.

 

With kind regards

Marlowe

[P_Marlowe]

Hi

 

answered your primary question, let me just comment

on the simulation you are trying to do

 

A) is pointless, you make the hand weaker, so the jack will

be a pos., and the only thing you find out is, how much

overvalued a jack is, assuming p has a certain number

of cards in the suit of the jack.

 

B) is a little bit more interesting, because now the HCP strength

is the same, and the question is now, if King becomes an Ace,

Aces being undervalued / jacks being overvalued, does this

still hold, if you assume, p has a certain number of cards in the

suit of the jack.

Similar if the jack goes and a Queen gets promoted to a King.

...

=> In my opinion you should break down the results of your simulation

into B1 (King gets promoted), B2 (Queen gets promoted), B3 (Jack gets

promoted) and B4 (a spot gets promoted).

 

Good luck.

 

With kind regards

Marlowe

[Cascade]

I did this simulation, which is more specialized that what you asked:

 

4-3 spade fit

 

15-17 HCP (with four spades) opposite 10 HCP (with three spades)

 

No eight card heart fit

 

No eleven card minor fit

 

No suit with four cards or fewer between the two hands

 

I generated 10000 hands and looked at dummy dummy results with the 15-17 hcp declaring 3NT.

 

The results surprised me a little.

 

Here are the average number of tricks taken for each number of HCP in spades in the three-card suit.

 

0 9.202893437

1 9.184325109

2 9.114208022

3 9.132590529

4 9.124835742

5 9.036968577

6 8.880866426

7 8.963636364

8 9

9 8.6

 

I used dealer.exe

 

Here is the coded condition:

 

spades(south)==4 and
spades(north)==3 and
hearts(north)+hearts(south)<=7 and
diamonds(north)+diamonds(south)<=10 and
diamonds(north)+diamonds(south)>=5 and
clubs(north)+clubs(south)<=10 and
clubs(north)+clubs(south)>=5 and
spades(north)+spades(south)>=5 and
hearts(north)+hearts(south)>=5 and
hcp(north)==10 and
hcp(south)>=15 and
hcp(south)<=17

[Tcyk]

That is very interesting Wayne. The decreasing number of tricks with increasing responder strength in spades is probably telling us that we should be playing in the 4-3 spade fit. I suppose you could run the simulation again with a spade contract instead of no trump. I don't have dealer. When I want to do a simulation, I cheat and look at what happened in the real world by running BridgeBrowser. I have not done so yet as I have just read your comments.

[TimG]

I did this simulation, which is more specialized that what you asked:

 

4-3 spade fit

 

15-17 HCP (with four spades) opposite 10 HCP (with three spades)

 

No eight card heart fit

 

No eleven card minor fit

 

No suit with four cards or fewer between the two hands

 

I generated 10000 hands and looked at dummy dummy results with the 15-17 hcp declaring 3NT.

 

The results surprised me a little.

 

Here are the average number of tricks taken for each number of HCP in spades in the three-card suit.

 

0 9.202893437

1 9.184325109

2 9.114208022

3 9.132590529

4 9.124835742

5 9.036968577

6 8.880866426

7 8.963636364

8 9

9 8.6

Doesn't this just tell us that the point is more useful in another suit, not necessarily how useful it is in the 43 spade fit?

[hrothgar]

Recently I've seen some members of the forum saying that Jxx in partner's 4 card suit is a good holding, I think exactly the contrary, so I would like to try a simualtion.

 

I haven't ever made a simulation, is it possible to make it compute a hand double dummy, and then compute the same hand A: removing the Jack for a small spot, B: removing the jack for a small spot but putting something worth 1 HCP on another suit.

 

If its possible please tell em with wich programs.

Hey there

 

I think that there are two reasonable choices for generating detailed simulations. Hans van Stavern's Dealer and Thomas Andrews Deal.

 

Using either program requires that you learn how to write simple scripts. If you have any experience with C this should be quite easy. I find Dealer easier to use, however, that is probably a reflection on my own coding profiency more than anything else.

 

Please keep in mind that the cardinal rule regarding simulations is Garbage In, Garbage Out..

 

I believe that simulations can be very valuable. However, I think that you need to do a a very good job framing the question you are trying to study.

 

In this case, you'll need to do the following:

 

Design a script that describes a set of hands that will play in a Moysian spade fit rather than an alternative strain. (This set of hands might vary dramatically based on what system is being use. For example, playing MOSCITO you'll play in lots of 4-3 fits. Playing a more standard system you'll play in a lot less). You might find it useful to look a specific auction like

 

1m - 1♠

2♠

 

You won't be immediately able to generalize your results, however, it will be easier to defend yourself against folks basing the methodology.

 

Once you're able to create an appropriate corpus of hands, I'd start by looking at two different pieces of information.

 

Too start with, I'd look at the probability denisty functions describing the expected number of tricks that any Moysian will take.

 

Next, I'd look at the PDF that describes the expected number of tricks that a Moyisan with Jxx in the short suit will take.

 

I'd try to see whether or not this is an

 

Average holding

An above average holding

A below average holding

 

Hrothgar

 

(Who is still feeling damn smug about scoring up +110 in a 3-3 heart fit on Saturday evening - For the record, I am feeling smug about the result AND the contract)

[TimG]

I didn't know this was about playing in a 43 trump fit with Jxx. But, here is what I have done.

 

Declarer = 4432 or 4333 with 4 spades and 14-16 HCP

Dummy = 3334 with 3 spades and 9-11 HCP

 

With 25 partnership HCP including J74 in spades: 8.561 tricks in NT.

With 25 partnership HCP including 874 in spades: 8.654 tricks in NT.

With 24 partnership HCP including 874 in spades: 8.164 tricks in NT.

 

[edit: each was done with a 1000 deal sample.]

 

This suggests to me that, given a hand with 874 opposite a 4-card suit, it is more valuable to add a point in another suit than to change the 874 to J74.

 

But, remember that the point added elsewhere could, in effect, be changing a Jack to a Queen, a Queen to a King or a King to an Ace. So, the point being added elsewhere is not the same as the point being added when we change 874 to J74 -- not all points are created equal no matter how ingrained the 4321 count has become. Rather than conclude that the point is more useful elsewhere, one might just as easily conclude that the simulation suggests the Jack is overvalued in the 4321 evaluation method.

 

I think it is going to take a much more detailed simulation to get an idea whether Jxx is "good" opposite partner's 4-card suit.

[helene_t]

I think it is going to take a much more detailed simulation to get an idea whether Jxx is "good" opposite partner's 4-card suit.

Yes. It begs the question "good compared to what?". If I have xxx in one suit and Jxx in another, I would rather have the jack opposite p's 4-card than opposite his void or 8-card. If I would rather have it opposite his 4-card than opposite his 5-card is less obvious. Maybe this specific problem should be analyzed: give opener a 5422 and responder two 3-card suits, xxx and Jxx, opposite opener's two long suits. Or maybe that wasn't the question ....

[TimG]

I would guess that the question being asked is something like this:

 

Pard opens 1S and I raise to 2S. He now makes a natural game try in hearts. Which do I prefer: 8 HCP with Jxx in hearts, or 8 HCP with xxx in hearts? Which should you be thinking: I have 8 HCP and my Jack is helping; or I have 8 HCP and my Jack is likely not pulling full weight?

 

But, then there were comments about a Moysian so clearly others are thinking of a different question.

[P_Marlowe]

I would guess that the question being asked is something like this:

 

Pard opens 1S and I raise to 2S. He now makes a natural game try in hearts. Which do I prefer: 8 HCP with Jxx in hearts, or 8 HCP with xxx in hearts? Which should you be thinking: I have 8 HCP and my Jack is helping; or I have 8 HCP and my Jack is likely not pulling full weight?

 

But, then there were comments about a Moysian so clearly others are thinking of a different question.

Hi,

 

Thanks, at least an example which illuminates, why the

question may be worth asking.

 

But the example also specifies the set you will need to

look at, a 1 level oppening bid facing a single raise.

 

And now p, makes a natural game try: What are the min.

req. for being able to make a natural game try in a given

suit.

 

With kind regards

Marlowe

[hrothgar]

I would guess that the question being asked is something like this:

 

Pard opens 1S and I raise to 2S.  He now makes a natural game try in hearts.  Which do I prefer: 8 HCP with Jxx in hearts, or 8 HCP with xxx in hearts?  Which should you be thinking: I have 8 HCP and my Jack is helping; or I have 8 HCP and my Jack is likely not pulling full weight?

 

But, then there were comments about a Moysian so clearly others are thinking of a different question.

Hi,

 

Thanks, at least an example which illuminates, why the

question may be worth asking.

 

But the example also specifies the set you will need to

look at, a 1 level oppening bid facing a single raise.

 

And now p, makes a natural game try: What are the min.

req. for being able to make a natural game try in a given

suit.

 

With kind regards

Marlowe

The original post did NOT state that this was a one level open opposite a single raise.

 

I number of people, myself included, have made assumptions about what hand types are being considered. However, only Fluffy knows for sure... (I'm not even sure if we're looking at situatins where Spades are trump)

 

As I already stated, its crucial to try to nail down precisely what you are trying to answer...

[P_Marlowe]

<snip>

As I already stated, its crucial to try to nail down precisely what you are trying to answer...

I agree.

 

And it may be helpful to have an idea, what the result

most likely will look like ...

 

If you do a simulation and the result is counter intuitiv go

back to START (without getting any money) and try again,

it is possible, but usually you have overlooked something.

 

With kind regards

Marlowe

[ASkolnick]

Cascade, first off. I love the dealer.exe program. It is so easy to manipulate and make shapes.

 

But for simulation, instead of dividing by HCP, you may want to use the suit quality function as well.

 

For example, I am sure J9x probably works better than J32 for ruffs as well.

[Fluffy]

J9x makes intra fineses double dummy, better stay away from it. J7x as minimum sounds better.

 

My idea at first was what Tim suggested, playing on a 5-3 fit, having a 4-3 fit outside.

 

But I could live to any hand where there is a 4-3 fit and we play in our best fit if there is one of 8 cards, or NT otherwise, seeing the impact of the Jack moving/promoting another honnor somewhere else.

[hrothgar]

J9x makes intra fineses double dummy, better stay away from it. J7x as minimum sounds better.

 

My idea at first was what Tim suggested, playing on a 5-3 fit, having a 4-3 fit outside.

 

But I could live to any hand where there is a 4-3 fit and we play in our best fit if there is one of 8 cards, or NT otherwise, seeing the impact of the Jack moving/promoting another honnor somewhere else.

In this case, I'd recommend the following type of construction:

 

Give Dealer a 1M opener

Give responder a raise to 2M

 

Require that dealer has 4 cards in a side suit and sufficient strength for a long suit game try (Presumably, the reason that you care whether Jxx is a "good" or a "bad" holding" is that opener showed a second suit)

 

Require that responder has three card support in that same suit

 

Compare the number of tricks that you take with a holding of J76 compared to 765. It's pretty easy to construct a hypothesis test to see whether the distributions differ from one another in a statistically significant way.

 

(Constraining responder's hand to hold a single raise should address some of the issues regarding shifting the position of honors as opposed to adding HCPs)

 

I recognize that this is a pretty specific construction, however, I think that tis better to use a simple constrained problem to nail down the methodology before generalizing the problem.