|
Numbr2intheWorld
|
02-22-2008, 01:20 AM
Post subject: simulating poker? (math project)
|
#1 (permalink)
|
|
Moderator
Join Date: Aug 2006
Posts: 3,561
|
|
So i have a friend who i'm helping on this project. He's trying to simulate a HU 5-card draw game to see the effects of variance in poker. I think it's a really interesting project.
Basically we laid out a design for a bot, writing out all the actions a player could go through in a limit HU 5-card draw game. By pitting 2 players who play exactly the same against eachother, we can see exactly how much variance there would be against two players of = skill level. Also we can tweak on players play to be better than the others (by a few simple adjustments) and find out how variance would work for a slight winner and a slight loser.
Thats all i really helped him with, but he called me up today saying that, although he thought he knew how to make a simulation of this, he actually could not. I told him i'd ask FTR about it.
So:
Do you guys know anyway to write a program to simulate this?
If there is, is it too complicated to do with a short time period?
any suggestions on other ways we might accomplish this?
|
|
|
Play for FREE and practice your game at...
Join the FTR Poker Forum to disable these banners and start posting!
|
|
gabe
|
|
Moderator
Join Date: Dec 2004
Location: trying to live
Posts: 7,964
|
|
im a computer science major and its not too complicated.
but theres already stuff to measure this, but i guess its cool if hes doing it for fun.
|
|
|
|
Irisheyes
|
|
4-of-a-Kind
Join Date: Jun 2005
Location: over there
Posts: 3,708
|
|
Everything in computer programming is done with i=0;i++;i<x. Thats all you need to know.
|
|
|
|
swiggidy
|
|
4-of-a-Kind
Join Date: Sep 2005
Location: Waiting in the shadows ...
Posts: 3,777
|
|
Quote:
|
Originally Posted by gabe
im a computer science major and its not too complicated.
but theres already stuff to measure this, but i guess its cool if hes doing it for fun.
|
If you have any programming experience it would be pretty simple. I don't think there is a bot simulation program, not sure that would be a good thing if there was (or would it??).
FWIW, I don't think it would be too hard either.
|
|
(\__/)
(='.'=)
(")_(")
|
|
|~|ypermegachi
|
|
4-of-a-Kind
Join Date: Aug 2004
Location: emo-kid
Posts: 3,580
|
|
here's some pseudocode for you...
X = {some range of hands before the draw}
Y = {some range of hands after the draw}
read cards
if hand is better than X
bet or call
else
fold
draw cards
if hand is better than Y
bet or call
else
fold
doomswitch code
|
|
|
|
shazbox
|
|
Straight
Join Date: Jan 2006
Location: Ottawa, Ontario
Posts: 170
|
|
Quote:
|
Originally Posted by gabe
im a computer science major and its not too complicated.
but theres already stuff to measure this, but i guess its cool if hes doing it for fun.
|
I musnt understand it right, it sounds pretty complicated to me.
I guess if you had all the rules they follow written out though it wouldn't be too difficult.
But making them actually play good would be hard. I guess you dont need them to play good though, just to play the same.
...Maybe i am just a really dumb comp sci major.
|
|
|
|
swiggidy
|
|
4-of-a-Kind
Join Date: Sep 2005
Location: Waiting in the shadows ...
Posts: 3,777
|
|
Quote:
|
Originally Posted by shazbox
Quote:
|
Originally Posted by gabe
im a computer science major and its not too complicated.
but theres already stuff to measure this, but i guess its cool if hes doing it for fun.
|
I musnt understand it right, it sounds pretty complicated to me.
I guess if you had all the rules they follow written out though it wouldn't be too difficult.
But making them actually play good would be hard. I guess you dont need them to play good though, just to play the same.
...Maybe i am just a really dumb comp sci major.
|
The way I interpreted the comment (because then it lines up with what I was thinking, haha) is that it would be easy to write a program that allowed you to do a simulation.
I would have a difficult time coming up with an algorithm the modeled decent play, that would be much easier for gabe. In the end, if they're playing the same I don't think it would matter too much. I would expect the graphs to look like a random walk. The thing you wouldn't get by modeling a coin flip is the stdev in BB/hand.
I guess if you came up with a sample distribution of how often the final pot is 2bb, 3bb, 4bb, etc, you could assume 50% chance of winning with a variable weight and that would have to be pretty close.
/rambling
|
|
(\__/)
(='.'=)
(")_(")
|
|
shazbox
|
|
Straight
Join Date: Jan 2006
Location: Ottawa, Ontario
Posts: 170
|
|
Ya it wouldn't be hard to just make a simulation.
|
|
|
|
spoonitnow
|
|
Straight Flush
Join Date: Sep 2005
Location: IRC Chat Room
Posts: 5,406
|
|
Greenstein has some C++ libraries on his site that evaluate 5 and 7 card poker hands. I've done a bit of work in the past with them when I was working on some short-stacking stuff and it's not too hard to deal with.
|
Quote:
|
Originally Posted by Ripptyde
I only have 2 simple rules when I am coaching a new student.
Rule # 1: don't ask questions
Rule # 2: don't ask questions
I have no interest in discussing strategy with a protege'. Your job is to remain quiet and listen. I have a very systematic approach that I will share with the right candidate and I promise that I will turn you into a force of nature and show you elements of the game of poker that you never knew existed.
|
|