simulation software suggestions?

[rbforster]

I like figuring out the "best" way to bid hands, especially in situations where it isn't clear. A nice recent example is where a garbage stayman approach with 4=4=1=4 leads to much better contracts than just passing a strong 1NT with a poor 5 HCP hand in simulations. I was quite surprised and that's the sign of a good experiment!

 

I'd like to explore this and similar types of simulations, and would like to know what software tools exist and what people here think of them. I understand that other people may have other preferences, but ideally I'm looking for something that

 

1. lets me flexibly select the hands to meet certain (hopefully arbitrary) conditions

 

2. lets me specify the lead and contract, and perform a double dummy analysis

 

3. lets me automate running simulations of many such hands and collect the results

 

Linux/unix OS, free, and open source would all be nice too, but I'm willing to pay for something that's flexible and fast for doing simulations. If you've used either hand generating software and/or double dummy software, what have you liked and have you been able to program it to solve the kinds of problems you wanted to? Thanks!

[rbforster]

found some discussion here

[Tcyk]

I like figuring out the "best" way to bid hands, especially in situations where it isn't clear. A nice recent example is where a garbage stayman approach with 4=4=1=4 leads to much better contracts than just passing a strong 1NT with a poor 5 HCP hand in simulations. I was quite surprised and that's the sign of a good experiment!

 

I'd like to explore this and similar types of simulations, and would like to know what software tools exist and what people here think of them. I understand that other people may have other preferences, but ideally I'm looking for something that

 

1. lets me flexibly select the hands to meet certain (hopefully arbitrary) conditions

 

2. lets me specify the lead and contract, and perform a double dummy analysis

 

3. lets me automate running simulations of many such hands and collect the results

 

Linux/unix OS, free, and open source would all be nice too, but I'm willing to pay for something that's flexible and fast for doing simulations. If you've used either hand generating software and/or double dummy software, what have you liked and have you been able to program it to solve the kinds of problems you wanted to? Thanks!

I have a suggestion that might meet your needs in a direction that you have not considered. BRidgeBrowser lets you examine hands that have been played on the Internet. Many millions of hands are available but its not free.

 

You can search for hands played by a particular player or partnership or even defenders. You can search by opening strength and/or shape. You can even include overcalls. You can search by final contracts like how often was 7NT bid and how often was it made.

 

The real advantage is that you can see what was done in the real world, not double dummy. I once set out to see what was the best 1NT opening point range. I searched some 20 million deals in 1 point increments from 8 to 20 HCP with balanced hands. The results were surprising (at least to me). 8 to 10 HCP had the best results in both MP and IMP scoring. The higher the opening range, the worse the results. This doesn't necessarily mean that mini-no trumps are the best method. It probably shows that the average opponent doesn't know how to deal with them.

 

I did things like searching to see which players made psychic opening bids and trying to learn my own short-comings. You get to see how all others bid and played the same hand as you.

 

Inquiry and Sfbp are associated with the program. I think Stephen wrote it. I am sure Ben helped collect data. I was an alpha and beta tester of the program. I can't come up with the web site off the top of my head but I am sure that a search for BRidgeBrowser will locate it.

 

Double-dummy analysis takes time. I use PS Bridge Hands which is freeware but requires GIB's double-dummy analyzer. It accepts hands written in PBN format which most programs support and gives the contract for all hands in all suits and no trump and the par result.

[hrothgar]

Right now I use Hans van Stavern's Dealer for almost everything.

 

Given that I am working for the MathWorks , I should probably figure out how to leverage MATLAB for this type of work.

[Rossoneri]

Right now I use Hans van Stavern's Dealer for almost everything.

 

Given that I am working for the MathWorks , I should probably figure out how to leverage MATLAB for this type of work.

An idea I had as well, but given my very limited experience with Matlab, I have no idea where to begin.

[ASkolnick]

I also used Hans Van Staverens dealer program which is great with its shape function. However, some of the random number generator I questioned.

 

Henk Uterwaal created an improved version of Hans Van Staverens dealer which includes functions such as suit quality, controls, and ability to create a PBN file. It runs off a C simulator:

 

Here is the link:

http://www.xs4all.nl/~henku/html/dealer/dealer.html

 

I have used this dealer with conjunction of Bridge Analyzer and deep finesse to come up with some theories. I created my own Access program to capture some data.

 

Whats nice about the dealer program is you create your own classes of information. You can also use other classes to build more classes, If you know C, it uses the basics of the language.

 

So you can build a definition of 1ZPM (1 Major in our system shows 31-35 ZP which has been defined as Controls+HCP+2 longest suits+(Long-short suit)). Once the definition has been created, it can be used in sequences.

 

The shape function is the most flexible I have seen of anything.

 

You can:

 

shape (south, 4531) Exact Shape

shape (south, 45xx) 4 Spades, 5 Hearts

shape (south, any 45xx) Any 4-5

 

and I believe in some PERL language he wrote an even more complex shape function.