How are poker bots raking online money?

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

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

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.

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

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

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.

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

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

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

@Mark
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

@Jake
That makes sense. Thanks for clearing that up.

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

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.

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

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