A Poker related final year CS project

sikes

Hi,

I know theres a load of people here who have studied computers or are very computer literate. Just wondering if anyone has an idea for a final year computer science project that would be poker related.

From the other thread, i was thinking i could write a program to simulate the tribecca/pokerstars etc tables and open X amount of them and analyise how the cards fall to see if they are "rigged" over a sample of millions of hands.

Another thing was a study of blinds and level intervals to see how they relate to starting stacks, and write a program that would dynamically change the blind levels to ensure a crapshoot doesnt evolve! though thats not particurally complex and more of statisitical project.

any help appreciated.

5starpool

hand histories are something that remain the companies property I believe and you would be on shaky ground there if you you could convince people to hand you all there HH's (and I doubt you could get enough to be statistically accurate even then).

sikes

i wouldnt be asking for peoples hand histories as there is no guarntee of their accuracy.

what i would do would be to write a program and let it open a hundred tables or so and "watch" the games. When there is a showdown record what happened into a database. Then after a couple of million hands, analyse them.

shepherd

How about something based on evolutionary principles. Has always interested me anyway.

Evolve heads-up strategies perhaps.

Probably much more work than than a final-year projets tho.

5starpool

sikes wrote:

i wouldnt be asking for peoples hand histories as there is no guarntee of their accuracy.

what i would do would be to write a program and let it open a hundred tables or so and "watch" the games. When there is a showdown record what happened into a database. Then after a couple of million hands, analyse them.

This is definitely something that most sites will not let you do afaik. Even PokerTracker is not able to datamine on most sites anymore. I think if are interested in doing something like this you would need to get their permission first. Good luck with it if you get it off the ground though.

ianmc38

Programme a winning poker bot. Now there's a challenge. You can get boards players to play it in 0.10c heads up games.

mewso

5starpool wrote:

This is definitely something that most sites will not let you do afaik. Even PokerTracker is not able to datamine on most sites anymore. I think if are interested in doing something like this you would need to get their permission first. Good luck with it if you get it off the ground though.

Datamining is not necessary. There are already programmes out there that "watch" the tables like Calculatem Pro and Poker Office. They need the software to be visible on the screen and from there they can record what cards are appearing etc. Not too sure how it's done tbh.

El Stuntman

Sikes,

best of luck with this (I won't even charge you for stealing my idea!!:) ).

just make sure I get some credit in your thesis

I can see it now: "Thanks to El Stuntman for his paranoid/delusional, consipracy theorist, drug-induced madness - he set me on the way".....

seriously though, sounds excellent - also sounds like you are much more qualified to do this than me...maybe ask karlh for permission to datamine PPP cash tables?

califano

How about doing a real project sikesy?.
This is like a lifer starting his first days incarceration and putting in for library detail!.

ChipLdr

Hi Sikes,

I am just coming to end of fourth year after completing a degree in ICT. I too wanted to do a poker related FYP. The problem i found was the lack of supervisors willing to take on my project.

I wanted to design a poker software program that allowed ppl to play poker against others using their mobile phones.It seemed that nobody wanted to know about my project and i stuggled to find a willing supervisor.

My advice to you is based on what you want to get out of this project? If your objective is soley to get a good grade(which most are) without caring too much about the topic then i reccommend biting the bullet and doing a project that has been offered but tailor it to suit your needs.

On the other hand if your goal is self satisfaction from implementing your own idea and a possible good grade then by all means go for it. But you must take into account the extra amount of work involved in implementing your own idea.

While picking a poker related project seems like great fun there are a few things to be wary of. Things like being assigned a clueless supervisor will severely hamper your progress if you get stuck.Hand Histories and the like will be foreign to them so this is an extra step(and a good bit of work ) to consider.

As the FYP gives the same reults as an exam i ended up picking a project offered that would get me a good grade but with the minimal amount of stress.

I'd advise weighing up the pros and cons before deciding. I didnt want to do a set project eitrher but i actually ended up having great fun implementing a P2P Spam + Bulk Mail removing program(nerdy i know).Trust me, a willing supervisor is ever so important.

My $0.02758 cents

sikes

How about doing a real project sikesy?.
This is like a lifer starting his first days incarceration and putting in for library detail!.

i have a problem in that i really dislike cs, and am definately in the wrong course, but its taken 3 years to realise this. so my final year project has to keep me interested and i have no desire doing any of the other final year projects, they are just boring so i hope to get one of the wild cards. just need an idea!!

you wouldnt have any would you!!

and i like your pun, no wait its not a pun...ok i have no idea what you call it but its good

sikes

Cheers for the abvice, ChipLdr. I suppose the grade is more important really. I have put my name down already for a neural network sort of project, predicting financial markets, which sounds interesting enough. If i cant think of a good idea and find a willing supervisor i will prob go with it.

our stats lecturer said he would sponsor FYP that have stats involved so i could ask him.

Poker & Pints

I have no idea how intricate a program you can create. But, I think it would be pretty cool of you could create a program(similar to Holdem Pro) which you could create different syles of players(agg, tight, etc) and have simulated games played. You would produce profit and loss results as a simple output. But more complex results would be how the dynamics of a game can change when different types of players enter and exit....i.e. a table full of tight players with 1 aggressive player, then 1 tight player leaves and another aggressive player takes his place. Then you can also chart all kinds of other data.

cardshark202

ianmc38 wrote:

Programme a winning poker bot. Now there's a challenge. You can get boards players to play it in 0.10c heads up games.

No. Thats my project. If someone else takes it I'm fúcked.

ArmaniJeanss

sikes wrote:

Hi,

I know theres a load of people here who have studied computers or are very computer literate. Just wondering if anyone has an idea for a final year computer science project that would be poker related.

Write your own person to person HoldEm game, programming the action/RNG etc. Give it the twist that half the screen is live Webcam shots of your opponents faces etc.

AJs (who has absolutely no idea if any of that is feasible, or hasnt been done already !)

ecksor

My housemate (artificial intelligence student) was mentioning a possible final year project being offered in their department based around AI and computer vision and trying to interpret your opponents facial expression looking for tells. I don't see it being particularly successful, but certainly interesting.

hotspur

I was pondering about 2 days ago the fact that there isn't any research on the impact of rake at small stakes poker tables. The fact of rake being acculumative in the sense that one doesn't merely pay 5% on net winnings but everytime one wins an amount of money irrespective of whether one has won and lost the same X money 50 times has an effect which I'm unsure of.

What I'm suggesting is a computer programme which analyses the monetary dynamic of small stakes online poker tables particularly with respect to the effect of rake. At low stakes where rake is likely to be actually 5% of the pots how much money does it take off the table in a defined session and what results are needed by players to "beat the rake". This is quite maths based but a programme which permitted comparison to actual experiences of players vis a vis beating or not beating the rake at low stakes poker would be original and very welcome, it's not a huge leap from some of the Pokertracker type programmes but certainly would have a different and more aggregate focus. You would have to work it around the acceptability of datamining of course.

Whatever you choose the best of luck to you.

strewelpeter

sikes wrote:

Cheers for the abvice, ChipLdr. I suppose the grade is more important really. I have put my name down already for a neural network sort of project, predicting financial markets, which sounds interesting enough. If i cant think of a good idea and find a willing supervisor i will prob go with it.

our stats lecturer said he would sponsor FYP that have stats involved so i could ask him.

A Neural Net ...predicting Financial Markets...

That is a poker related project.
Substitute the outcome of a hand for a player based on observation and or Hand Histories of particular scenarios.
Don't get too caught up in in trying to program a AI bot you willl find that there is more than you will be able to handle in applying it fully.
Just take one small aspect (preflop action as an indicator of final outcome or something like that) and try to do that well.

OTOH if you are let do a stats based project ...then what Hotspur said. And good luck too.

DeVore

Here's an evolution of that idea. I did my final year project on Neural Networks (wrote one that learned to play tic tac toe, I watched Wargames the film too much). This is something I would love to see done.

The neural network should be fairly simple considering that you have fairly constrained inputs but sufficiently complex to benefit from the Neural approach. It need only be able to play a heads up game against another copy of itself.

Once thats done, spawn a large number of copies of it into a "space" and let them play each other. Record wins and losses and at some point "kill" the lowest 10% to introduce selection pressure on the population.
Next include reproduction by having two "players" merge their networks to create a new, third, player. This replenishes your population from the 10% cull. If you want to get fancy you can introduce minor occasional copying errors to represent mutations and miscopies ala genetics.

This second part of the project is generally called Artificial Life though its known by other names. Read anything by Dawkins for more info. Theres a lot of interesting work done in this area.

Hopefully you would find an optimum playing strategy for that population but even if not it would be interesting to see if the population split into loose aggressives, tight players, loonies etc etc.

I'd be very interested in this project, enough to offer any help I could.

DeV.

sikes

cheers dev, sounds like a great idea. will do some research on this and a couple of others and will let you all know how i get on. hopefully i will get to do something like this.

Wallko

lol Dev and the others, dont think iv understood a single word used in this tread!!

DeVore

sikes wrote:

cheers dev, sounds like a great idea. will do some research on this and a couple of others and will let you all know how i get on. hopefully i will get to do something like this.

During the WSOP last year they were holding the World Series of Bots (cant recall what it was called). I meant to go but woke up hungover and didnt
You should look into that too, as its like Robot Wars but for poker.

DeV.