BBO Discussion Forums: generating specific hands using dealer and bbo dealer.php - BBO Discussion Forums

Jump to content

Page 1 of 1

generating specific hands using dealer and bbo dealer.php problems with instructions and generating

#1 User is offline   bobtodd 

  • Pip
  • Group: Members
  • Posts: 5
  • Joined: 2020-June-11

Posted 2020-June-16, 10:55

Hi- I tried user BBO's practise to generate hands but that wasn't very successful perhaps because my constraints were too tight. I install dealer on Ubuntu Linux. $Revision: 1.24 $
$Date: 2003/08/05 19:53:04 $ $Author: henk $. I got some decent results but still some difficulties. Found dealer.php on BBO and pasted the rseults from Linux dealer and ran.
Good results.

I wanted to generate inverted minors in a wek NT structure (12-14). First test was opening hand was 15-17 4432 or 4333 opposite 5+D and 12+HCP plus a little extra shape. This is what Linus dealer generated.
generate 500000
produce 20
dealer north
condition (shape(north,4333 + 4432) and hcp(north)>=15 && hcp(north) <=17) and (shape(south,3343 + 3361 + 3163 +1363 + 3154 +1354 +1264 +2164 +2155 +1255) && (hcp(south)>=12 and hcp(south)<=18))
action printoneline

sample output was: (as expected .. there was more data)
n A982.K832.K97.AK e K5.Q975.63.Q8754 s JT4.A64.AQ84.J93 w Q763.JT.JT52.T62
n AKQ2.J943.943.AJ e T8763.K87.J82.92 s J.A5.AK765.KQ863 w 954.QT62.QT.T754
n K542.AJ74.A64.K2 e 86.T82.Q92.A8753 s AQ3.Q.JT853.QJT9 w JT97.K9653.K7.64
n AK32.AKJ.T42.J54 e JT7.84.AQ73.T973 s 5.65.KJ965.AKQ82 w Q9864.QT9732.8.6

Here is the problem.

When I pasted the source into dealer.php on BBO that also worked.
When I went to Bidding practise and pasted the same source into "Advanced"
I get output but not very good output.
I don't think I can control the "generate" command.

Also I specified "dealer north" as above - but dealer kept rotating.
In order to generate the less frequent distributions such as 3154 1264 etc, I think generating lots of hands is important. Maybe BBOs practise interface is to constrained.

Can anybody provide some guidance? the Deal program is very nice as is the Pre-processor dpp but they are a little hard to grasp and get used to.
an anybody tell me what is wrong with specifying "dealer north" in source/Advanced and why my distributions are so flat.
Thanks
0

#2 User is offline   paulg 

  • PipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 4,562
  • Joined: 2003-April-26
  • Gender:Male
  • Location:Scottish Borders

Posted 2020-June-16, 11:25

The BBO implementation is a subset of the Dealer program. All you need for the BBO bidding table is

condition (shape(north,4333 + 4432) and hcp(north)>=15 && hcp(north) <=17) and (shape(south,3343 + 3361 + 3163 +1363 + 3154 +1354 +1264 +2164 +2155 +1255) && (hcp(south)>=12 and hcp(south)<=18))

You can constrain the dealer using the drop down menu at the top of the Deal Source->Advanced tab.

You can not generate a specific number of hands, you just the get the first ones that match.
The Beer Card

I don't work for BBO and any advice is based on my BBO experience over the decades
0

#3 User is online   hrothgar 

  • PipPipPipPipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 14,771
  • Joined: 2003-February-13
  • Gender:Male
  • Location:Natick, MA
  • Interests:Travel
    Cooking
    Brewing
    Hiking

Posted 2020-June-16, 11:27

One quick note:

I suspect that you want shape(north, any 4333 + any 4432)
Alderaan delenda est
0

#4 User is offline   pescetom 

  • PipPipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 3,640
  • Joined: 2014-February-18
  • Gender:Male
  • Location:Italy

Posted 2020-June-16, 12:59

If I may repeat one of my previous complaints, how to use dealer is too important and current to be hidden under "Full Dislosure and Dealer" when Full Disclosure is known to be a deprecated legacy. Probably best to duplicate and split.
1

#5 User is offline   bobtodd 

  • Pip
  • Group: Members
  • Posts: 5
  • Joined: 2020-June-11

Posted 2020-June-18, 17:53

View Postpaulg, on 2020-June-16, 11:25, said:

The BBO implementation is a subset of the Dealer program. All you need for the BBO bidding table is

condition (shape(north,4333 + 4432) and hcp(north)>=15 && hcp(north) <=17) and (shape(south,3343 + 3361 + 3163 +1363 + 3154 +1354 +1264 +2164 +2155 +1255) && (hcp(south)>=12 and hcp(south)<=18))

You can constrain the dealer using the drop down menu at the top of the Deal Source->Advanced tab.

You can not generate a specific number of hands, you just the get the first ones that match.


Hi,
Thanks for the reply.

Yes - I did constrain by using the advanced tab. That lead to more questions. Can you use constraints + advanced. How do you add multiple constraints and verify that they are being applied. I had some strange results when I used Constraint 1 and constraint 2. Also sometimes when i edited constraint 2, constraint 1 went away. It would be really nice to see some instructions that about setting constraints.

The only way I could sensibly use 2 constraints was to edit the data in advanced and type in the 2nd set of constraints.

When I set set up the rules using dealer. I copied those into advanced. I also added a line "dealer north". That did work and I assume I have to add that to the string Condition (dealer north && (shape ...... )) to get north to deal as well as setting up other constraints in Advanced.

Bob
0

#6 User is offline   bobtodd 

  • Pip
  • Group: Members
  • Posts: 5
  • Joined: 2020-June-11

Posted 2020-June-18, 17:58

View Posthrothgar, on 2020-June-16, 11:27, said:

One quick note:

I suspect that you want shape(north, any 4333 + any 4432)


Hi,
Good thought except I actually wanted to generate hands that held exactly 4S 4H 3D and 2C opposite more shapely hands.

Because of the issues I was having generating useful hands I started restricting 1 hand type for N and different types for south. I would then generate other hands for North such as exactly 3343 and then separately deal any flat hand for North.

It seems of the constraint is too general you don't get enough interesting hands.

Using the linux program dealer and using generate and produce, you get much better hands.

Bob
0

#7 User is offline   bobtodd 

  • Pip
  • Group: Members
  • Posts: 5
  • Joined: 2020-June-11

Posted 2020-June-18, 17:59

View Postpescetom, on 2020-June-16, 12:59, said:

If I may repeat one of my previous complaints, how to use dealer is too important and current to be hidden under "Full Dislosure and Dealer" when Full Disclosure is known to be a deprecated legacy. Probably best to duplicate and split.


I also agree with this ;-)
0

#8 User is offline   bobtodd 

  • Pip
  • Group: Members
  • Posts: 5
  • Joined: 2020-June-11

Posted 2020-June-18, 17:59

View Postpescetom, on 2020-June-16, 12:59, said:

If I may repeat one of my previous complaints, how to use dealer is too important and current to be hidden under "Full Dislosure and Dealer" when Full Disclosure is known to be a deprecated legacy. Probably best to duplicate and split.


I also agree with this ;-)
0

#9 User is offline   armantt2k 

  • Pip
  • Group: Members
  • Posts: 6
  • Joined: 2020-January-06
  • Gender:Male

Posted 2020-July-10, 10:57

I find the Dealer in BBO Teaching Table doesn't appear to provide truly randomized hands.

Specifically, when I put in just a simple HCP <9 constraint on both the EW robots, we get seem to see an EXTREME number (way beyond what you would expect randomized) of 65xx or 66xx hands, as well as many 7, 8 card suited hands in the robots. This leads to lots interference bidding by the robots - preempts, Michaels, UNT, etc. Also, our NS hands seem VERY often to have extremely unbalanced HCPs (6v19, 25v3, etc)- again, multiple times per 15-20 boards and EVERY session, way more often than I think should occur if Dealer was truly randomizing.

Has anyone else notice this? Also, does anyone know if perhaps there is a SEED number I need to add to change this?

Thanks for any thoughts by those more experienced with Dealer - I've only used the BBO interface. I'm running Windows, not Linux - is there a Dealer for Windows version?
0

#10 User is offline   pilowsky 

  • pilowsky
  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 1,108
  • Joined: 2019-October-04
  • Gender:Male
  • Location:Australia
  • Interests:Writing, learning about life and bridge from others, Poodles, helping Poodles write

Posted 2020-July-10, 15:59

Just found this site. Looks like it does everything! http://playbridge.com/index.php
Can someone show me how to take their lin files and put them into BBO so that I can use them for bidding and play practice?
non est deus ex machina; även maskiner behöver lite kärlek; N'écris jamais une lettre et n'en détruis jamais une.
0

#11 User is offline   smerriman 

  • PipPipPipPipPipPip
  • Group: Advanced Members
  • Posts: 1,802
  • Joined: 2014-March-15
  • Gender:Male

Posted 2020-July-10, 17:37

View Postarmantt2k, on 2020-July-10, 10:57, said:

I find the Dealer in BBO Teaching Table doesn't appear to provide truly randomized hands.

Specifically, when I put in just a simple HCP <9 constraint on both the EW robots, we get seem to see an EXTREME number (way beyond what you would expect randomized) of 65xx or 66xx hands, as well as many 7, 8 card suited hands in the robots. This leads to lots interference bidding by the robots - preempts, Michaels, UNT, etc. Also, our NS hands seem VERY often to have extremely unbalanced HCPs (6v19, 25v3, etc)- again, multiple times per 15-20 boards and EVERY session, way more often than I think should occur if Dealer was truly randomizing.

Has anyone else notice this? Also, does anyone know if perhaps there is a SEED number I need to add to change this?

Nope; either you have something wrong with the conditions, or, more likely, either "what you think should occur" is incorrect, or it's simply a case of confirmation bias.

The chance one of the robots has a 7 card suit or 6-5 or better is about 1 in 10; over 20 deals it wouldn't be unusual at all even to see this happening 4 times (15%).

Given your constraints, there's a 1 in 4 chance one of N/S has 18 or more points (and about the same that there is a 9 point difference between the hands). Even 13 point gaps you should regularly see more than once a session.

I just dealt a set of hands under those conditions at a bidding table and there was nothing unexpected about those two situations.

Many people seem to misunderstand that when you consider a bunch of hands, the chances of seeing 'rare' events, even multiple times, increases very quickly - and considerably more so when you start looking for them in multiple aspects, like shape and HCP. You will then never fail to notice them, even though they are expected.
0

Share this topic:


Page 1 of 1


Fast Reply

  

1 User(s) are reading this topic
0 members, 1 guests, 0 anonymous users