In this article, I will discuss methods to avoid having your Poker Robot detected. (Books shouldn`t be worried, as the next article will show you how to crush the robots.)
From what I`ve read/heard, there are three basic types of Poker Robots:
1) Programs that merely assist a human player with advice. 2) Programs that play automatically while you sleep. 3) Programs that play in collusion with each other.
For Type 1 robots, there is no advice I can give (or is needed) for the player. You are the one typing in bets and the computer is serving the same function as a very good book, table of stats or friend giving advice. (Depending upon the implementation, the book may be able to crush this robot, but very tough to do. More on that next week!)
For Type 2 robots, the books are looking for your patterns, since computer programs behave differently than human beings. As was mentioned in a recent article, robots can play several games at once and don`t need bathroom breaks. So, Rule 1: don`t run your robots 24/7. Run them a few hours at a time.
If you can manage different dial-up accounts (different ISP and IP addresses), try having more than one account at a poker site and rotating between accounts.
Rule 2: Don`t be stupid and run the next one two minutes after the last one stopped. That would constitute a pattern. Now, you might want to run each account at similar times of the day (one profile is a night owl, the other works during the day, etc.)
Rule 3: Vary the amount of time you play and when you start. Don`t play from exactly 12:01 am to 4:23 am each time. Set up a general time frame and roll some dice to pick start times and length of play. Dice rolls are more random than anything you "think" is random.
The poker sites look for inhumanly quick behavior. Rule 4: Don`t use a robot that always responds in one second to every play. For you bot-programmers, throw in a little random number generator from 5 to 15 seconds for the response time.
I don`t know if any programs have variations on strategy (playing loose or tight), but this would be a good variable for the player to be able to adjust. Rule 5: If you can adjust strategy and run more than one account, run each with slightly different strategy. If possible, adjust your own strategy a bit over time. No human plays exactly the same tightness every time.
For Type 3 robots, the poker sites look for the same players at the same tables and making bets that would be unusual for players if they had no knowledge of the other player`s cards. Too much cooperation will yield patterns.
The programmer has to be careful here. Colluding robots should probably try to predetermine which players are the strongest and always "lean" that way, but not so much that other robots don`t win on occasion. You might want to program some outward tells into the robots so when you beat them (and then get caught), you can argue that you were picking up on the "tells", rather than knowing the cards. While a "tell" might weaken your robot against a human player slightly, it probably won`t overcome the advantage of collusion.
Rather than all the robots converging on a single table in heads-up play, a better collusion strategy may be to have them all enter a tournament. In a tournament, they will all be randomly assigned tables and playing in non- collusion mode. However, if they advance in the tournament, they will be randomly thrown together and can collude a bit. (I`m assuming here that the type 3 robots can play as type 2 when no partners are present.)
Next Week: Poker Bots, Part 2: How to Crush the Robots