View Single Post
  #438  
Old 05-09-2007, 07:04 PM
LB_001 LB_001 is offline
Senior Member
 
Join Date: Oct 2005
Posts: 191
Default Re: NL Bots on Full Tilt

"Over the next couple days, I started playing again and would observe the bots’ tendencies.
-First thing I noticed is simply how much they play. They played non-stop for several hours almost every day. They put in more hands than anyone else at 1/2. They are probably some of the highest volume players on the site at any game at any stakes.
-Next, I noticed that the three bots never ever ever ever sat at the same table. For players that play so often, this would obviously be highly unlikely even over short periods of time. Yet, they played every day for months without sitting at the same tables.
-All three accounts would sit down and get up at the same times.
-Their betsizing was the same. They would virtually always make a pf raise to the size of the pot, reraise to 3x the original raise, and always make a continuation bet of around 75% of the pot (regardless of the number of players or the flop texture).
-The bots would “think” excessively long times to make every single decision, even trivial ones. One hand: bot limps utg, fish minraises, six calls, back around to bot who takes 15 seconds before finally deciding to call getting like a million to one. Also, they would cbet almost every single flop yet they would still think for 10-15 seconds before they made the cbet, every single time.
-Occasionally, the timing went the other way. In one hand, I flopped TPNK HU against one of the bots. I c/c bets from the bot on flop and turn. On the river, I made an absurd blocking bet of $5 into a $60 pot just to see how the bot would react. The bot insta insta insta folds. The micro-second I clicked bet, the bot folded. Even if they had nothing, a person would still have to think for half a second to consider whether they should bluffraise my silly looking bet. After that hand I decided there wasn’t the slightest doubt, I was definitely playing against a computer.
-They never chatted (in the first several days I watched). This would change."



I think it's pretty obvious that all the accounts are operated by just 1 player since they never sit together and play at similar times, but I'm not convinced they're bot accounts. The owner is obviously at the computer a lot or all of the time since he is chatting and now seems to be playing differently. I'm not sure why the argument of he's looser in late position or folding to reraises or floats indicates he's a bot unless he's folding to extremely tiny bets consistently or something. Many players always bet around 75% of the pot as well. Some of the timing tells might apply but according to OP it was only over a few day period and the insta-folds once or twice could be just the guy clicking the check/fold button ahead of time. The guy could also be using a program to automate preflop actions since the actions are usually decided by hand strength/position and not much thinking involved so this could explain why it takes the same amount of time for him to act. Someone mentioned there's a program that automates preflop that some people use already, I guess it's up for debate whether that's fair or not.

Obviously something a little strange is going on since one guy is using 3 accounts instead of just 1. Not sure if there's a table limit or something at full tilt or if he's just a little nervous about getting an account busted or something. If they were 100% bots the guy could just leave them on excessive hours every day instead of just "several hours a day". Even if they were 100% computer controlled the guy wanted to monitor them rather frequently and was scared to leave them by themselves for long periods of time for fear of being exploited. I think another good indication of a bot would be how the player rebuys (i.e. does he instantly top off his stack no matter how small below the max buyin it drops or if there's a delay or a certain stack % threshold before rebuying).

edit: just saw Nation's post but I'm still not convinced that three different people would have almost exactly the same stats. It just seems extremely unlikely that they'd play so similarly over very large sample size. Also it's very strange that one account would log off and right after one of the other ones would log on unless they shared accounts, or that all three would play at the exact same time, I suppose it is possible though just a very strange setup.
Reply With Quote