Rise of the Robots

The_Daddy_H

Anybody here ever try to write a poker bot?

I'm curious about this subject and may consider giving it a try.

Mr. Flibble

The_Daddy_H wrote: »

Anybody here ever try to write a poker bot?

I'm curious about this subject and may consider giving it a try.

I made one that killed play money games on Tribeca by pushing with TT or better and folding the rest. I left it overnight and it turned 1k into 40k. Then the Tribeca software changed the next day and that was the end of that. If you search google you will find a website/forum dedicated to poker bots. I can't remember the name of it right now.

opr

http://pokerai.org/openholdem.html

Think this is one Mr Fibble means.

Opr

The_Daddy_H

Thanks for the links, some interesting stuff there. seems like there is quite a large community developing these things.

It doesnt seem all that difficult to program a reasonable poker bot or at least one capable of playing the ABC poker required to beat an averge $1/$2 player or below, I find it hard to believe that the lower limit games arent already infested.

I mean essentially 99% of the most important decisions at the basic poker level is simply a question of a range V range algorithm. A quick and dirty calculation of a range shouldnt be all that difficult in most cases.

Goodluck2me

surely its against the rules?

ianmc38

goodluck2me wrote: »

surely its against the rules?

lol. why would you think that?

The_Daddy_H

goodluck2me wrote: »

surely its against the rules?

Whats against the rules?

Pulsar Eagle

The_Daddy_H wrote: »

Whats against the rules?

bots

Fremen

goodluck2me wrote: »

surely its against the rules?

It would probably be forbidden by most poker sites' EULA, yeah. It's not easy to stop without being extremely invasive, though. You'd have to monitor all the processes running on the client's computer. People don't like that.

opr

Fremen wrote: »

It would probably be forbidden by most poker sites' EULA, yeah. It's not easy to stop without being extremely invasive, though. You'd have to monitor all the processes running on the client's computer. People don't like that.

Even with this i would have thought it wouldn't be that hard for someone with enough knowledge to build one of these to go undected.

Like running the bots remotely so the bot software is entirely undetectable on the client machine.

Edit - i actually think that as time goes on this will become one of the biggest problems online poker faces.

Opr

pokerbunnybear

what is a poker bot?

ArmaniJeanss

pokerbunnybear wrote: »

what is a poker bot?

A computerprogram which plays poker for you whilst you do something more productive.

jhegarty

If you run poker software in a virual machine window (vmware/viruta pc) and the poker software on host computer it would be 100% undetectable...

hotspur

Fremen wrote: »

It would probably be forbidden by most poker sites' EULA, yeah. It's not easy to stop without being extremely invasive, though. You'd have to monitor all the processes running on the client's computer. People don't like that.

Poker sites do exactly that. You don't even want to have the winholdem site up when you are playing poker.

Tar.Aldarion

A really good bot could be written, I just don't want to be the one programming it!

The_Daddy_H

Well I never said I was planning on running it on a poker site to make money. My interest is purely academic.

As far as circumventing poker room security measure is concerned, this is really quite trivial.

Besides in a worse case scenario you could train a troop of monkeys to hit a key based on what the poker bot says to them.

DubTony

The_Daddy_H wrote: »

Well I never said I was planning on running it on a poker site to make money. My interest is purely academic.

:eek:

Besides in a worse case scenario you could train a troop of monkeys to hit a key based on what the poker bot says to them.

That's already been done. Haven't you seem any of phantom_lord's hand histories?

dvdfan

For any new sites wouldnt it be great if they could get their hand on a few bots, they could instantly have liquidity at all hours of the day and the money they lose would be saved on paying rake. Imagine 4-5 bots at every table, they could colude to ensure they beat the real players so if im Bot 3 and ive 99

Send Query Bot1;Bot2;Bot4;Bot 6;
Var WhatyouGot;
If Bot1 ! Bot2 ! Bot4 ! Bot6 = 9 SET WhatyouGot = True;
If WhatyouGot = true THEN FOLD ELSE CALL

Maybe its already happening, Maybe all them conspiracy theories have merit?

opr

Bots that can see your hole cards would be better imo

Opr

krux

Program a good one and betfair will probably consider sponsorship.

hotspur

dvdfan wrote: »

Maybe its already happening, Maybe all them conspiracy theories have merit?

There have been poker sites in the past which used bots at limit hold em for liquidity reasons. Also we pretty much know that user created bots have been used at high stakes HU limit hold em, especially at Ultimate Bet.

Mr. Flibble

The_Daddy_H, here is a decent article on the matter: http://www.codingthewheel.com/archives/how-i-built-a-working-poker-bot

The_Daddy_H

Mr. Flibble wrote: »

The_Daddy_H, here is a decent article on the matter: http://www.codingthewheel.com/archives/how-i-built-a-working-poker-bot

Thanks for the link. Yes, I found this article on reddit yesterday and was going to post it. Its very interesting indeed. Seems like the bot problem is somewhat overblown at present but rising rapidly.

It seems like the bulk of the code is available for free on the net, its the customizing of the bots poker strategy thats up to the you. This is really the aspect that I'm most interested in. It still requires a considerable amount of coding and the bot will probably never be smarter than its creator, at least without some crazy machine learning techniques which I'd imagine would be very difficult to implement for NL holdem.

Mr. Flibble

Have you read much about them since first posting? Do you know if people have bots that keep a history of the players and adjust to how their playing style? By using a PT database for example?

eagle eye

Tar.Aldarion

The_Daddy_H wrote: »

Thanks for the link. Yes, I found this article on reddit yesterday and was going to post it. Its very interesting indeed. Seems like the bot problem is somewhat overblown at present but rising rapidly.

It seems like the bulk of the code is available for free on the net, its the customizing of the bots poker strategy thats up to the you. This is really the aspect that I'm most interested in. It still requires a considerable amount of coding and the bot will probably never be smarter than its creator, at least without some crazy machine learning techniques which I'd imagine would be very difficult to implement for NL holdem.

I think a bot could be 'smarter' than it's creator. It would be very hard though, adding pt type data would be a great start. Genetic algorithms might take a while to come up with a good one though...

Drakar

Mr. Flibble wrote: »

Have you read much about them since first posting? Do you know if people have bots that keep a history of the players and adjust to how their playing style? By using a PT database for example?

There's at least one out there that shares its hole cards with other bot players, but I havent heard of one which plays differently against different players depending on their stats (I think it's too early in the development cycle for this, but it's undoubtably coming).

Camone

Just saw this thread, for my final year project I wrote a mathematical model for playing optimally in Razz, pretty happy with it, but the calculations involved are too long and arduous to do in any way but a computer program.

So, if there are any competent computer programmers out there whod b interested in working on such a project and think they'd b able to convert the formulae into a program, pop me a PM

Mr. Flibble

Do your equations ever bluff (serious question)?

RoundTower

how could they not?

camone if you really have the skeleton of perfect strategy in razz I think a lot of people would be interested in seeing that published.

hotspur

Mr. Flibble wrote: »

Do your equations ever bluff (serious question)?

If it didn't it would be about the worst razz bot imaginable.

Glowingmind

Bots are bad. Don't use bots.

Camone

Mr. Flibble wrote: »

Do your equations ever bluff (serious question)?

at the moment the model just gives a probability reading on each street that we will be ahead of our opponent by the river, given his distribution, cards out etc; as regards bluffing, i will only be able to do it against players i have a lot of previous hand info from, monitoring things like how often a player will fold given certain combinations of cards for both of us, which usually make him lose interest in a hand (we might find for a certain player, 80% of the time he will fold on 4th street if he is faced wit bets on 3rd and 4th and 3rd and 4th bring him two higher cards). i dont use pokertracker or any such software so im not really sure the extent of the info i wud b able to get. i would then assign a weighting to both reading depending on the opponents propensity to call. I havent begun work on this side of the project yet, plan to look into it after the exams
the model is by no means perfect and definitely has limitations at the moment but i feel that wit a bit of work there are definitely some games which it could beat on the reg.

The_Daddy_H

Mr. Flibble wrote: »

Have you read much about them since first posting? Do you know if people have bots that keep a history of the players and adjust to how their playing style? By using a PT database for example?

I done a little bit of reading since. Seems like its a very involved project to get a bot up and running even if more of the nuts and bolts are already in place. Obviously nobody wants to sell a good poker bot.

I think poker tracker has been implimented to work in conjuction as well, given the trouble of making a bot to begin with I'm sure adding some info from pokertracker wouldnt be too much extra.

I dont know for sure but intuitively the thought of using the genetic algorithm with NL texas holdem make me very queasy.

RoundTower

Camone wrote: »

at the moment the model just gives a probability reading on each street that we will be ahead of our opponent by the river, given his distribution, cards out etc;

that doesn't sound very interesting at all tbh, how can you say this is "a mathematical model for playing optimally in razz" any more than pokerstove is a model for playing optimally in hold'em?

what does your model say you should do when it is folded to you in last position before the bring in and you have (KK)4 and the guy has a Q up? Should you bluff here against an unknown player or only if you have a few hundred hands on him showing his tendency to defend the bringin?

Drakar

The bigger bots out there have their default rules set for limit as this is much easier for bots to play. I'd imagine there are quite a few of them running on sites all the time as there's little the sites can do to stop them (leaving aside the debate of whether they'd want to).

Mr. Flibble

Here's a follow up article to the one posted above.
http://www.codingthewheel.com/archives/how-i-built-a-working-online-poker-bot-2

sneakyaces

slansky created a poker program and offered 1/2 million to any1 tat cud beat it

Mr. Flibble

And part 3:
http://www.codingthewheel.com/archives/how-i-built-a-working-online-poker-bot-3

Tight Ted

Online Poker is doomed. The only one who has a chance is Vanessa Rousso, she knows all the math!

http://www.rawvegas.tv/watch.php?vID=73c96c93bb5968799e0a099a8e4de2