I am really interested in using pokerbots and I would like to know how advanced is poker bot technology nowadays. And which are the top poker bots at the moment?
OK let me tell you about bots, where they are extremely strong and where they are weak.
I have played a lot of Omaha on-line and I was playing against the best known bots on PokerStars and had a winning record. When this was acknowledged by Stars and refunds were issued, I didn't get anything. The most damning thing is Stars have by far the best software in the industry, were TOLD about the bots and it still took them a year to end the rake-machine. So bots are there even on a site with security that can emulate your system and analyse all your plays in real-time. There is a perverse incentive for poker sites to allow bots as they accrue rake faster than human players - they don't really care who wins as long as they make more money.
How good are they? They can be statistically unbeatable in HU FLHE or any FL game where equity calculations aren't skewed by large bluffs and raises. The main edge bots have over human players is that software can accurately select the chosen ranges your opponents statistically hold, live, at the tables.
The problem of getting data on all opponents (10 billion+ hands in your HUD) is overcome by software that was once legal. Personally I always considered it cheating, never used it and now major sites prohibit such software. I don't even use HUDs any more as they bore me.
So here's the thing. If each human played a rule-based GTO (Game Theory Optimal) strategy to the best of their abilities and didn't deviate from that, the bots would win. Any human with the knowledge and skills to play close to GTO is a pro who is likely playing 5+ tables simultaneously. The calculations for perfect GTO based on opponent history and ranges are beyond a human being but child's play for a bot loaded with hand histories.
The top poker bots right now are unknown by definition. A top bot not only has to accrue data and perform all the poker calculations but most importantly, it must avoid detection. A lot of 3rd-party software is now banned on major sites and they DO check but it's not possible to stop all bot-related software - e.g. screen scrapes, virtual systems and other off-line ranging and calculation tools (e.g. Excel) that don't interact with the poker client software.
Even in PLO, an advanced game, a handful of bots made $millions. There are good discussions about the PLO bots on 2+2:
It's also known that bots were beating NLHE MTTs, which are massively more complicated than cash game PLO. Another ring of cheats was broken when a group of professional players from eastern europe used sophisticated bots to handle the deep-stack parts (the start) of MTTs and then humans took over towards the end of MTTs when bots could no longer make optimal decisions.
If you want to know how to beat/avoid bots it's relatively simple:
1) Play NL/PL games with as many players as possible. I no longer play 6-max or hyper-turbos because those games are mathematically solvable for bots. A human playing a few tables but with nice software can easily do the same thing as bots in 6-max HT. If you follow the high volume regs on-line you can see exactly what strategies they employ. It's so obvious to me their actions are the result of computation. That's also a weakness for them so:
2) Be random. Don't always do the same things with the same hands. If you alter your range and how you play that range then no software can ever accurately range you. If you mix up your levels of aggression and do randomly stupid/brilliant things (like most players do by accident) this will confuse bots/software calculations
3) Exploit the bots. The bots are playing GTO based on ranges, bet sizes etc. So a bot would CBet Flop, bluff turn and river etc at the optimal rate. The problem for the bots is their lack of dynamism. Any pro player will pick up on the patterns used by each bot and if you have faith in what you are doing, you can destroy bots with crazy raises and shoves that destroy their calculated equity and ICM considerations.
Say the bot is on the button, makes a standard pre-flop raise with A T and the flop is A J J. You flat call pre with A Q and flat call all the inevitable small bot CBets. You will scoop at showdown, fine, but what happens if you 3-bet/shove the flop, turn or river? Well depending on your individual profile (your VPIP, PFR, Call|PFR etc) the bot will either snap call or snap fold. Once you understand how the bot is reacting to you and the rules it must follow, a top pro can out-play a GTO bot with increasing ease as the number of players at the table and game complexity increases.
There is no such thing as "Artificial Intelligence" and there won't be for at least 2 decades. A bot plays only according to the rules that govern it and can never learn, only be manually updated.
Having said that, if I had the desire I'm sure I could make an unbelievably potent bot. The main reason is I know how the individual regs and pros like to play poker, their unique styles. So it's fairly easy for me to add my knowledge to an already existing bot to specifically target a group of players (say the 1,000 top Omaha players, whom I know individually).
Of course if bots collude, especially in non-holdem games, they have the potential to be unbeatable both statistically and by any other means. Their edge is so large from playing 8-12 cards instead of just 4 that any half-decent bot would destroy the best players in the world without question, in any format.
Man, todays poker AI technologies are very advanced. There are a lot of rumors by the obvious reasons - bots earn a lot of money to their holders. Moreover, there are huge farms that make money on rooms. It works vice versa so rooms use bots to support "rake machines".
BTW, seeking guys that use bots to make money is very difficult task since they are hiding. But they are always open to scaling their business so you can spend some time searching info about poker AI and find people that are interested in attracting new partners.
I recently watched an interview with Jungleman and he said bots are being used more. He thinks it has been solved for limit.
No limit is much more difficult to solve.
He thinks they are more common in sit n go.
protected by Community♦ Jul 3 '18 at 5:44
Thank you for your interest in this question.
Because it has attracted low-quality or spam answers that had to be removed, posting an answer now requires 10 reputation on this site (the association bonus does not count).
Would you like to answer one of these unanswered questions instead?