Poker Botter Discusses How He Won $30,000 on ACR
We all know bots exist in online poker, though to the extent that they are a problem is up for debate. Clearly, poker botters stay in the shadows; their operations run contrary to site rules, so if they were caught, they would have their money taken and accounts banned and possibly even face legal ramifications, depending on their jurisdiction. One botter, though, recently discussed his exploits on Two Plus Two, explaining (while still remaining anonymous) how he profited about $30,000 with his bot in six months.
The botter, who appropriately chose the screen name “themadbotter” for the message forum, said that he had run a bot on America’s Card Room (ACR), a member of the Winning Poker Network (WPN) for about six months before cashing out. He initially revealed some of the software and other various technical solutions he used to allow the bot to interact with the poker room, but they were rightly edited out by moderators. He explained that actually getting a working bot up and running is easy; the more complicated part is winning:
The more time consuming endeavor is getting the bot to play well. Plug-n-play bots generally come preloaded with profiles that–at best–are capable of playing slightly winning poker at the lowest limits or freerolls. There are forums and marketplaces where a botter can buy better profiles, but these can’t play very profitably above 10NL on most sites. The best solution is to write your own profile. This used to be the barrier to botting a couple of years ago but with the proliferation of PPL (or oPPL), it now takes only a couple of hours to learn the syntax of coding your bot profile. Making it play exceptionally well is still extremely time-consuming and it requires a lot of trial runs and hand history reviews. For me, it took me about a week to write a profile that played well enough to beat 10NL and about a month of reworking that profile to beat 50nl at a solid clip. Over the course of the next few months, I steadily improved the bot’s performance based on manual review of hand histories and results.
He went on to explain that he put a lot of work into making sure his bot acted like a real person so as to go undetected. He didn’t have it play around the clock, it switched tables and joined waiting lists, he programmed in misclicks, had it type generic comments in the chat box, it didn’t play the same hands the same way all the time, so on and so forth.
When asked, themadbotter gave some examples of what people could look for when trying to detect bots. For instance, basic bots are often programmed to type in a bet, calculated as a percentage of the pot, whereas his bots used the 1/3 pot, 1/2 pot, 2/3 pot and similar buttons. He said that if you see an opponent make a bet that is very close to, but doesn’t exactly match, what the bet would be by pressing the pre-set buttons, there is a chance that player is a bot. This is because, say, the 2/3 pot button may make a slightly different bet because of rounding than a 66 percent bet – a real person generally wouldn’t calculate 66 percent of the bot, but would instead just hit the 2/3 pot button.
On the flip side, if a player is making very exact calculations correctly all of the time, he could also be a bot. “A more glaring misstep,” themadbotter wrote, “is when a player bets something like 60% of the pot…correctly every time…despite the the number of players or pot-size on whatever street. Think about whether you would go through the effort as a human to figure out what 75% or 60% or 65% of a random pot on the turn might be every single time and manually type it in when you are playing 6+ tables.”
Our botter also noted that a sign of a bot is if it is playing lots of short sessions around the clock. An incorrect assumption is that playing for many hours straight is a tell-tale sign of a bot, so novice botters will set their up to play for a couple hours, then take a break for a few hours, etc. The problem is, to play enough hands to make botting worth it, it would have to play at very odd times using this schedule. A real person would not play a half-dozen one to two hour sessions per day with naps in between.
He made it clear, though, that just because someone displays these patterns does not necessarily mean they are a bot.
One interesting takeaway is that themadbotter feels that most poker rooms are perfectly fine with having bots on their tables unless enough players make a fuss about them:
In general most sites will look the other way for botters unless the botter garners a massive number of complaints from other players. Bots and poker sites have a mutually beneficial relationship. Bots are the most low-maintenance, high-value customers that most sites have. Bot-operators will never complain about cashout times, won’t e-mail support over and over about mundane issues, won’t complain about other players, won’t be ill-mannered, etc. Good bot-operators will remain low-key and continue to churn out rake for sites and fill up the tables to boost player #’s.
He also does not feel guilty about botting:
The answer to that is a resounding NO. I suppose I don’t consider myself a cheater. I didn’t share any hand-histories or real-time data and I was not part of a bot ring. I probably put in more hours studying sessions, opponents, and general gameplay than 90% of profitable players. Botting is easy, botting well is not. It’s not as if the bot suddenly turned me into a winning poker player. I was a good cash game reg before (on a different site) and a big part of the reason why I was able to construct a good profile was because I was a solid player with close to 2M hands of 6-max playing experience.
In the end, themadbotter quit because it was taking up too much of his time. He was apparently very proactive in analyzing the play of his bot every day and making the necessary tweaks to the code that it was almost like a second job. He is now moving on to a “more lucrative career” so he just won’t have the time anymore. He claims he wasn’t necessarily in it at first for the money, but rather as more of technical challenge, but of course once his bot started doing well, he enjoyed seeing it bring in some profit, so he kept it going for an extra few months.