Bots in Online Poker

By Deane Barker on September 24, 2004

Are poker ‘bots’ raking online pots?: How would you even enforce this? There’s no way to touch their client machine, so how could you ever know what they’re running there?

Concern is growing in online chat rooms and news groups devoted to poker that sophisticated card-playing robots — known as “bots” in the nomenclature of the Web — are being used on commercial gambling sites to fleece newcomers, the strategy-impaired and maybe even above-average players.

Say they did make some detection device so that you had to have an actual human sitting in front of the machine. What’s to stop two people from playing — one to tap the keys and be the human, and the other to just enter what’s been dealt into a separate computer running a bot or analysis program.

And what about card counting in online Blackjack? Strikes me that bots were born for this.



  1. The whole point of an online poker bot is to make money without human intervention. It’s fairly easy to play well enough to beat low limit online poker. The bot’s appeal is it can do it 24/7 on many sites at once.

    Online blackjack sites shuffle the deck between each hand. Card counting isn’t effective in that situation.

  2. Say they did make some detection device so that you had to have an actual human sitting in front of the machine

    Hmm, and what sort of device would that be then ? Perhaps a trojan that you have to download before playing that switches on your webcam and lets the robot croupier and ‘droid security guy watch you while you play ;)

  3. Actually they would use a program that detected patterns of play that would most likely be play from a bot and not a human. This could get very tricky to program around but may still be possible. Supposedly some casinos are already using software like this and I am sure that innocent people are being kicked out too but I doubt the casinos care. The same type of thing happens with blackjack in RL casinos. If they detect certain patterns of play that label you as a counter, you are outta there whether or not you are actually a counter.

  4. I know for a fact there are programs that detect bots at sites . For one thing you can track response time, if they play only certain cards, whether they type in comments, and other things that are proprietary information so I can’t share them. As for the blackjack comment., it’s just like at most casinos now a days they use a continuous shuffler like ShuffleKing. The other thing is that is very hard to create a successful bot and right now it is impossible to create one for any game other than limit hold’em.

  5. Randomizing is what computers do well. So response times can be easily programmed to be randomized. I play a lot online and many players turn off chat for many reasons. They do not want it to affect their play. They are sick of the bad beat talk. They are in several games at once and chat is a distraction.

    The key point missing in this thread is that most bots can be changed from day to day or hour by hour to adjust ones play to be more aggressive or less aggressive. So it is not like all users are using the same formulas and the same play. Each person using a bot will be very different from each other using a bot. This makes detection even harder and practically impossible. Remember good players are like robots and play consistently the same way over and over again. With the use of two computers, simple checks for bot programs running on the same computer as the online poker software is running is not good enough.

    The real threat is that these bot programs allow for the ease of sharing cards and teams of players working together. Can you imagine 2,3, or 7 players at one table sharing what cards they have with each other and the bots take that information in their decision making. Bots and sharing cards are the biggest threats to online poker and its survival. Most poker rooms take a blind eye to these practices and are not equipped to deal with these issues. Frankly the attitude of most online poker sites is one of “as long as we are getting the drop (money) what does it matter”.

    Here is an example of how one person can be 9 players out of 10 player at one table. He has 9 computers in his room and he uses one local connection to the Internet. Then he uses 8 modems to call dial-up accounts (long distance) that are located in 8 different areas of the country. With 9 different poker accounts under 9 different names and funds coming from 9 different bank accounts in those areas where the dial-up accounts are located. The reason for different dial-up accounts is that then they are given unique IP address and the online poker site will see that they are located in 9 different areas. Ok so the sites have detection software to see that these players when they play together all at the same table. So instead of 9 poker accounts he has 40 accounts so that he can make it look random and not be detected. Why would someone go to all this trouble? Plan and simple, this is big money. I personally see no way that a poker site can ever guarantee that the above things are not occurring and that a live and unique player is playing at each spot on a table as long as computers are involved. Any technology that comes out will always be countered by other technologies by those who have the potential to gain. I see once these issues are exposed more in the mainstream media (news) then we will see a decline in the popularity of online poker sites and even their demise.

  6. we all now that bots are around if you search, google you will find bots availble for free!! demo only of course, but the point is there are bots out there. that are kept secret making money for the owner, you can build your own robot just type, (make your own poker bot) in google and the tools are there!!

  7. You can never totally find bots, someone who does not leave the table for 5 days and then automatically posts a blind everytime someone comes to the table is a little obvious. Survival of the fittest, if you can’t take the bots playing then leave the site, and if you are good enough to beat one then you will rake in some easy money.

  8. Hi,

    I’m a programer and also a poker player. I began with poker bots about 6 years ago. First bot was simple and it played only a preflop strategy. Maybe “bot” is not appropriate name for such simple program, but I will use “bot” word further here for all my programs. First bot had a table of playable cards under 3 main conditions. Number of bets before him (or me), position and holding hand. I used graphic and chatbox interface for spying in game progress. Most of today poker bots also spy graphic interface rather than network communication. If hands whare not playeble “bot” clicked on “Fold” button, otherwise he played me loud sound thru speakers so I could get to computer and play only good hands. Poker site for witch this bot was made had a maybe 300-400 online players in that time, so my type of play was not profitable because most players knew I’m playing only “legit” hands and I also couldn’t track their play and style, because I was away most of the time. After that I didn’t used any bots. I tried first comercial poker bot “WinHoldem”, but the way it was designed it was quite useles. By this I mean programing bot’s “brain”. About 2 years ago I noticed site witch had willson software holdem bots plugin for “poki academy”. I made interface to communicate with this plugin and poker room software, so bot actualy played the whole game like willson’s poker software does witch was leading commercioal poker software few years ago. I will tell you how this bot ended and how lastest bot works in my next post about latest poker bots. Here is just few things about comments before me.

    Custom bots that are not known to public are very hard to detect by poker room, because they don’t have informations how bot works. A good bot should include following features witch most of advanced bots have:

    • Changes agressivnes and tightnes factor a little bit, so play is not same all the time.
    • Bluffing.
    • own mouse driver with smart moving, delays and clicking, so there is no way poker room could detect that some other program is controling mouse.
    • Advanced math for bot’s brain such uses poki bot or maybe neural network would also do the job.
    • Not let it run 24/7

    There are also many other featuers that bot should have to avoid detection.

    I could write at least 10 pages about poker bots here, but it will take me a lot of time.

    Why poker room should even bother about bots if public doesn’t know. They also give rake as players does. Rake is profit for poker room.

  9. Hello NotMe, I’m also making a poker bot. Can you please give me more info about what you know to avoid the detection of a bot by a poker room? What do you mean by “own mouse driver with smart moving”. Can we use the user32 functions to click. Is it necessary to move the pointer before clicking?

    “There are also many other featuers that bot should have to avoid detection”

    Can you talk more about that? You can write me an email to

    Thanks a lot. Maybe will we exchange some techniques.


  10. I have had bots running online for over a year without detection. I am sure there are many more however every bot programmer I have come across has lost because he knew nothing about playing good poker. Creating the interface is easy but creating a bot that wins consistantly is beyond what most people have the time for.

  11. Hello Pistachio,

    On what AI techniques did you base your bot? Stuff like probability triple?

    Thanks for the input, cheers, tom

  12. Hello,

    By AI I think you mean opponet modeling? If so I have no AI but have hard coded a rather complicated strategy that took about a year to debug. The only way to debug a poker bot is by putting it into action and correcting all the bad folds or missbets. Many bot runners I know test out their bots by playing other bots which is totally the wrong approach. You must lose money in order to win later.

    I wrote a bot that tracked all the players and kept a database on every bet they ever made but I found it was not needed or accurate enough (due to not enough hands seen) to warrent a change in strategy. I noticed it would not often play against the same players and that a “generic” strategy was fine.. I think its overkill for most limits but may be needed in higher limits in which you see the same players often.

    Also..everybody’s idea on opponet modeling is way way off. They all track the wrong data. I dunno who came up with some of those ideas but they didnt play poker for a living thats for sure. I have a method that I know is superior to what is out there and I dont even use it. I have no idea what “probability triple” is but im sure a poker player did not invent it. Also..and this is key..many times you are up against two or three different “models” in one hand..then what? You either have a hand you can call or bet with or you dont. You cant win every hand but many bot programmers try to. They see that they folded the winner then decide that their model is no good and think they need a better AI.. This is impossible. I dont care if the biggest fish in the game bet into me when I have 55 and the board is KJ6. I am folding. I dont care if he called 72o to the river the last hand.

  13. You said creating the interface is easy, would you be able to tell me where I might be able to find some information on creating an interface for a poker bot to play online? Thanks.

  14. Hey POKO

    where can u download a standard bot. Can you customize an interface for me if i pay a bit of $$??

  15. I am one of those members of the public who frequents online poker sites consistently, and am not at all appalled by this but intrigued. I know my way fairly well around a computer, so if anyone wants to talk to me about creating/detecting bots I would very much appreciate their time. I don’t necessarily want to cheat at poker, but just even the playing field a little.

  16. There are many bots you can download already but they will not win much, if any, as they rely soley on math. They pretty much play the nuts so any dummy can do that. One is called Inspector Helper. There are several made around the Online Holdem Inspector odds program. They are pretty limited as they just play one table and do not move around. They have some real issues so they are pretty stupid as bots go. They will get crushed in any limit above .50/1

    If you want to create your own bot or interface your gonna have to learn C++ as the main hand evaluator that everybody uses is written in that (written by the Poki guys). That is the nucleus of all bots around right now. The interface to the poker site is just checks on whether various controls on the poker game window are active. Getting your hole cards is easy too. On most of the sites you will have to use image recoginition for that which will take 2-3 hours per site to get running right. Everybodys software is different so making a bot that runs on all sites will take lots of time. Remember that this is the easy part. Getting your bot to win and play strategy correctly could take months.

  17. hi pistacho your wrong, the fast c++ hand evaluator that everybody use is an open-source project called pokersource on source forge and was writted a long time ago , the poki guys make a java adaption that can call the native pokersource handevaluator. anyway if someone here is good in java (or native c++) and want a good poker framework that can connect with poker client msg at

  18. Choc,

    Oh right..not the Poki guys..I dunno what I was thinking. I use that evaluator as well for a rough estimate of what to do. I saw that you also post to the Poki group. How is your project coming along?

  19. I’ve spent the last few monthes working on a pokerbot for using Autoit3. I’ve just know finished the interface which works wonderful with the pstars client. I’m starting the very very long project of programming it to play, again in autoit3.

    It isn’t nearly as powerful or widespread as C++, but I’ve certainly made some progress, and I think that it is definately do-able for the low level tables at least. As for concerns about being caught, my interface is about as impeccable as they come. It can switch tables, limits, seats, AND accounts according to a set of user defined rules. It automatically selects 1 or about 10 different randomized input engines, I am very confidant there is no way pokerstars would be able detect it based on its mouseclicks.

    I’m not going to settle for a ‘static’ playing strategy either. It will take a long time to make(especially in a non-C environment like autoit). But its worth it because of the ability to interface with the totally graphical client pstars uses. for everyone out there that wants to learn more about writing pokerbots, feel free to send me an email at AT* I’d be happy to point you in a good healthy starting direction.

  20. where do u start? I dont know anything about C++, but my friends do. Is there software one can download, or a good website with resources? This is more for casual use/testing

  21. You don’t have to learn C++ or any specific language in order to create a bot that can interface with an online poker game. I wrote Inspector Helper (now only $10!) in VB.Net. I had to use a few API calls to do the screen scrapes and to position the Poker windows and such, but the rest is in plain old VB.Net.

    There is another application known as Online Poker Inspector (or OPI for short) that can “read” the situation and give a suggestion based on many important factors. You can adjust the settings (they call it a “profile”) and tell it what to do in almost any situation.

    It has some short comings. For example it should consider the pot odds in every situation, but it doesn’t. And it can not adjust its play to the type of players at the table. But if you know how to use it, it can still make money.

    You need to set up a profile that has strict starting hands, more strict than the ‘Solid’ profile that it comes with. Then when you do play a hand it needs to be aggressive. I find that a 2/4 table works best because anything less than that and the other players will call too many of your raises.

    My InspectoHelper simply automates OPI, but you still need to check on it every ½ hour or so. This is because if someone figures out your pattern then they can wipe the floor with you. Also, if there are too many aggressive players (players that raise pre-flop) at the table, then the bot will not do well.

    In the long term you have to switch your strategy (profile) every so often or else other players’ Poker Tracker software will give them an advantage over you.

  22. Joe,

    I find it hard to believe that Inspector Helper makes money over the long term. If its anything like the other OPI bot called PokerBotPro then it does not even know if the pot is raised or how many people are in the pot etc..Have I been bet into? Is nobody in the pot? what is my position? I cannot even fathom how a bot can operate without this knowledge. The other issue with OPI is that it blows a gasket on many playable hands post flop. OPI cannot semi bluff raise, steal or many other tactics that will yield quite a bit of money in the long run. I am going to guess your bot will check quite often when it should bet. Like it will not be able to bet thru a flush or paired board. Anybody can play the nut hands and win..the real bread and butter for a bot is correct play of weaker hands. This will make the difference between breaking even and winning.

    For $10 bucks your program is a real value and its definetly worth it however. I would think it would be great for clearning bonuses.

  23. I don’t think a programmer ought to make a poker bot unless he’s proven to himself that he himself plays profitable poker. That way, he can judge his AI by watching it play. If a programmer cannot notice problems from direct observation, he has no choice but to let it run 10,000 to 20,000 hands and let the financial results tell him there’s a problem! Obviously, that is not practical.

    I play profitable poker. I am a programmer and I have created and currently run profitable poker bots.

    It is very hard to program the AI to play profitably. It takes a lot of time. Sometimes the bot loses a lot of money and you begin to doubt whether you really know how to program a money making bot. This is exactly the same issues a poker player faces during long bad luck streaks. Any good programer with no poker playing experience is in for a very rough time.

  24. Hello,

    I totally agree with you. It took me months and months to get it right. I had to watch it make bad folds or miss bets and then correct the mistakes one by one. There are so many unique situations in poker that it was a much bigger task then I had initially thought. To this day I still find situations that I had not thought of until I actually see it. Like I mentioned before..the only way to debug a poker bot is to watch it lose money until all the mistakes are fixed. Luckily its not too tough to get it to break even.

    To be honest..every poker bot programmer I have come across over the last year and a half has given up entirely. It sounds easy but its not.

    It sounds like you play as many hands a month as I do (20k+). I would be interested in comparing notes on winrates etc if your up to it.

  25. For all of you programmers out there, we are looking to test out current bots that will enable players to easily enter their own system of when to bet, check and fold under as many conditions as possible. The bot simply needs to be able to play by itself, only one table at a time and must be able to be used on at least some of the online casinos for testing. It does not require any AI at this time. Future funding is available for further research. Please contact us at:

    Thanks! Poker Research

  26. Party has implemented a new anti bot mod. Other people have had issues with it but I have not seen it yet. A dialog pops up and asks you to type the letter/number combo you see in a graphic. I believe it gives you 30-90 seconds and if no response it deals you out of the game. I have not heard of any account freezes so far. This is probably just the start of things to come and will get tougher in months to come.

  27. Can You Beat A Poker Bot?

    Over the past few weeks I’ve read a lot of concern from the online poker community about the existence of poker bots. For those of you unfamilar with the term I’ll try to explain exactly what …

  28. PP caught my commercial bot. Be very careful. They seized my account and will not let me back in.

  29. Hi guys.

    Ive been reading these posts and reviews of bots. I am looking around for a program to use to create a bot. I was almost going to buy winholdem, but decided to to because of all the programming and advanced knowledge you have to know. I am looking to program a bot using AutoIt, but still have really no idea of what im doing. If someone knows or has experience in this field, i would appreciate a little help. Maybe if you could send me a link to a webpage for some assistance or a place where i can download software to help me.

    Thanks, Lou

  30. Concerned:

    Any commercial bot is a problem as they know what to look for. They also use screenshots and process attempts to help identify bots. This is why you must really program your own bot so they have no idea what to look for. I have not had any issues and my bot plays over 25k hands a month.


    AutoIt is easy to use. I would just read up all you can on that and get AutoIt source code snippets so you can learn faster.

Comments are closed. If you have something you really want to say, tweet @gadgetopia.