How are poker bots raking online money?

1 Star2 Stars3 Stars4 Stars5 Stars (3 votes, average: 3.33 out of 5)
Loading ... Loading ...

Jeffrey | Poker Articles, Poker Rules, Poker Strategy

Submitted by Thomas, this article belongs to the Poker Strategy series.

Sophisticated card-playing robots – known as “poker bots” – have now been playing online for years and the trend is accelerating. The increasing activity on poker bots forums and the growing interest generated by the annual poker bot competition (organized by the University of Alberta) are prime indicators of this trend.

Of course, online poker rooms have been fighting bots since the beginning as it hurts their
revenues but, detecting bots is becoming more challenging every day as bots improve their
counter-detection measures. While Full Tilt, one of the leading online poker rooms, recently
successfully detected a network of bots (and redistributed the seized bankrolls, thousands of
dollars, to the fooled human opponents), it only shows the tip of the iceberg.

This increasing presence of online poker bots proves, if still needed, that it is a very lucrative
business to be in. But how are poker bots raking online money? In other words, how come poker
bots are so successful against human players?
Thomas Kessler, founder of My Poker Coach (a free Poker coaching service based on a
leading bot), explains what makes poker bots so strong:
In the long run, poker bots are winning against humans and this can easily be explained by a
number of bots’ intrinsic properties:

• Analytical power: bots have, already today, a massive amount of analytical power and it
is going to further double every two years as computers become more powerful. Our bot,
for instance, simulates ~5 million situations at the Flop in less than a second to perfectly
assess the current hand strength and the possible outcomes. Something a human player
will obviously never match.

• Lifelong memory: can you remember how the opponent sitting next to you last month was
playing? Bots have a memory that will never fail. They can remember all their opponent’s
previous actions and playing style. In our case, we store all the observed moves (even if
our bot isn’t in the hand) in a large database and, next time we play against the same
opponent – even in 3 years, we already know his playing style.

• Discipline: most humans are open to emotions and are sometimes looking for adrenaline
in a game. Bots are disciplined and will, for instance, never go on a tilt. Discipline is key in
poker and always pays off in the long run.

• Game richness: there are many great poker strategy books out there but most players
remember less than 20% of it and they often over-use the few strategies they remember.
Bots can learn a large set of strategies and only apply them when most appropriate.

• Patience: who likes to fold 80% of his hands without seeing the Flop? Bots will only play
hands that should be played, even if they’ll have to spend most of their time watching
opponents playing (and collecting information on their playing style).

When asked what human players can do to defeat bots, Thomas says: to increase your chances
against bots you should definitely train yourself with them to learn their play pattern
and you should frequently change your online ID so that bots can’t rely on their historic database when playing against you.
So, next time you play online, ask yourself if you’re ready to play against bots!

Created by voluntary Artificial Intelligence developers, MyPokerCoach provides a free
online poker coach that has recently proven to be stronger than Poker Academy Pro 2 (the leading
commercial poker software).
The coach will provide play advices in a fictive online Texas Limit Hold’em game against 4 to 9 bots
(using play money) and give detailed explanations on the advices. It can directly be launched using
the button “Launch coach” on the MyPokerCoach site as no software installation is needed (it only
requires Flash version 10 to run in your internet browser).

Similar Posts:

Tags: , ,

12 Comments to How are poker bots raking online money?

Mark W.
May 13, 2009

The stuff from the University at Alberta is almost certainly never implemented on the internet. I guess you can’t rule it out, nobody ever would have ever thought a team of MIT students would take on Vegas, but guys like this are pretty smart, and its not really in the nature of academia to use their stuff for cheating at online poker. I think to paraphrase Bryce Paradis’ words… none of the people who are interested in using Polaris to cheat online would ever be able to implement it. Even if they could get their hands on the software.

I think Maybe in 3-5 years the bots you see online will come close to the level of Polaris, but these are only for limit games and I have doubts that many readers of this blog are seriously grinding it out in the midstakes limit games. IF you can’t beat a bot in a game of PLO, you should quit poker.

Mark W.’s last blog post..Some Terrible Weekend Sports Picks

May 14, 2009

So your saying that you can not be sure if you are playing against a real person online. But poker use bluff and many other techniques that a bot is not capable of doing. Even chess programs still can not beat the grandmasters and emotions are not so much a part of chess as they are in Poker.

Rose CARE’s last blog post..Rose Care or The Perfect Rose Garden

Paypal Poker Pro
May 14, 2009

From a mathematical point of view it would be possible to “train” a bot to be able to make money at limit tables, but limit is all about odds, especially at the lower stakes where bluffs will very rarely, if ever, work.

You can train a bot, but you could also train yourself to beat a bot playing mathematically “perfect” poker.

Poker Reviews
May 20, 2009

I would assume PPP is on the right page with the bots being able to make some money on low stake fixed limit tables because the game is very mathematical. I would like to think that most poker players would be able to beat a bot though since a bot can’t perform everything a person can.

Poker Reviews’s last blog post..Expekt Poker Signs Scotty Nguyen

May 21, 2009

I think bots have the potential of being superior at low level games, but not in high stakes games where meta game considerations dominate the decisions.

Mark’s last blog post..Personality Deficient

May 21, 2009

One one hand, it is clear that bot can or will be able to beat most players for the reasons described in this great article. On the other hand, top players will find patterns in them and exploit them, until they cannot. Or course ultimately bots will be stronger than humans, this is true for ANYTHING.
Anyway, from a more practical standpoint, whatever stealth techniques are used, a bot is still a bot, and poker rooms CAN detect them. Using highly advanced statistical pattern recognition techniques, but this is the end of my comment.

June 2, 2009

I think bots can be detected by poker sites using special software as they display a fixed playing pattern.

This is the case in chess where it can easily
be detected if you use a software chess program at home to beat your opponent. Then you get disqualified very fast.

Nobi’s last blog post..Sicilian Defence – get your counterchance

June 3, 2009

Hi Nobi

Thanks for you comment! I have a hard time picturing how a piece of software should be able to distinguish between a bot and a very disciplined / predictable poker player. Wouldn’t some players be falsely accused of being bots?


Mark’s last blog post..Playing Match-ups in Daily Fantasy Baseball Contests

June 6, 2009

One of the ways you can look for bots is to look for players who play identically. It is very unlikely for a group of human players to have exactly the same patterns over fx. 50k hands, so if you find someone who does, you have some prime candidates to be using a bot, and you focus your attention on testing those players.

Jake’s last blog post..Poker Datamining.NET

June 6, 2009

That makes sense. Thanks for clearing that up.

Mark’s last blog post..NY Giants Game Challenge update

June 17, 2009

The comments contradict the article. The article claims bots gather statistics on the opponents (like, how often does it fold to a raise on the turn, or how often is a river bet a bluff) and adjusts their playing style. Old bots like Winholdem follow static rules and will end up with the same stats, but those are easily beated. New bots adjust their strategy, if they win with preflop raises and turn/river bets, but lose money on calling other player’s flop bets, then the software will adjust and call less bets on the flop. This way also the statistics will vary as much as a real player, as their tactics change as a response to other real players.

A typical college kid that thinks he’s a pro and multitable plays 6-8 tables, he is probably following a set of rules in his head, like to fold 80% of lousy hands and to only play decent hands, then after the flop to fold whenever he has hit nothing. Because he plays so many tables, he’ll probably always play the same patterns. This way he slowly grinds a minimal profit, 8-12 hours a day. These tight multitablers usually make 2 BB’s (big blinds) an hour, which at $0.50/$1.00 tables is $2 per hour per table, $16 for 8 tables, $160 a day.

So what’s the diffence between such a multitabler and a guy who puts his tactics into a script and lets the software read the cards, execute the script to get either fold/call/raise and then have the PC press the buttons ? Nothing, except the bot user can run this while sleeping and do something useful during the day.
The only difference with the multitabler is that the multi guy sits at the PC, makes the same decisions in his head and uses the mouse to press the buttons. But the end result is the same, they play the same game.

June 17, 2009

Hi John

First of all thanks for your comment. Do I understand correctly that what you are saying is that it is difficult or even close to impossible to detect bots and that low stakes grinders would be better off programming a bot?

Best regards


Mark’s last blog post..All in poker radio show 10-Jonathan Little provides WSOP updates

Leave a comment

Special promotions

Latest Poker Strategy Articles

Contact PokerBankrollBlog

Questions? Concerns? Comments? Contact us! We'll return your email within 12 hours.