I haven't got a chance to listen to last nights show yet so maybe you already talked about this, but it doesn't look like anyone posted about this yet...
http://forumserver.twoplustwo.com/28...r-wpn-1619689/
Printable View
I haven't got a chance to listen to last nights show yet so maybe you already talked about this, but it doesn't look like anyone posted about this yet...
http://forumserver.twoplustwo.com/28...r-wpn-1619689/
Interesting. No, I hadn't seen this yet.
Here it is, in case 2+2 deletes it:
Quote:
I have decided to retire from botting and focus my energy on something more meaningful. I have carefully timed this post after steadily withdrawing all funds from ACR. I hope to shed some light on the current economy of botting and to help people understand the relationship between botting and poker sites. Over the past 6 months, my bot has played 500k+ hands on ACR mostly at 50-100NL. With various promos and bonuses factored in, it has generated around $30k of profit. I can't give much more details on results because, despite the fact that I don't have any money tied up in the network, WPN still has all my personal information and I would prefer to remain anonymous.
Currently, there are quite a few viable solutions for prospective botters to help connect the bot engine to the poker room. **** and **** accomplish the task well, but they have some technical limitations; **** is a open-source project that is better for custom-building a bot for a specific system and poker room. Plug-n-play systems generally require minimal technical expertise whereas **** and similar frameworks will require intermediate programming experience. In general, bots in 2016 all come equipped (or can be equipped) with stealth technology that will remove any overt indications of the software running during a session; they will generally show up as some nondescript process running in the background on task manager.
Finding or building a bot is the easy part. 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.
If you construct a good bot, it is impossible for other players to suspect the bot. For example, in addition to the stealth, my bot would randomly sit out and take breaks every couple of hours, never played for more than 6 hours at a time, frequently switched tables and joined waiting lists (with the aid of a hopper software), misclicked every so often, typed short comments in the chatbox on rare occasions, and had randomness built into the playing profile. It wouldn't always play the same hand the same way, wouldn't always play a missed flop the same way, wouldn't always valuebet the same amount, etc., and would use the half-pot/2/3 pot instead of typing in weird bet amounts. Around month 3 is when the bot really took off, after I worked tirelessly to integrate a Poker Tracker database with the bot profile. Depending on VPIP, aggression, cbet%, fold to cbet%, etc., and overall results of a particular opponent, the bot had close to 15 different 'branches' of play. It would play a nit much differently than it would a lag, a fish differently than a rock. It would exploit players based on tendencies: e.g. those who folds to 3bets too often, those who 3bet light OOP, those who folded to positional cbets unless they had top pair+, etc.
50nl was the bread and butter, and 100nl was profitable to a smaller degree. I never attempted 200nl because the reg pool thins out a bit and the overall better caliber of players made me hesitant to throw a bot in there: 1) because it would take constant reconfiguration to adjust to the regs adjusting to the bot, and 2) because the small pool of better regs might discover the bot's existence much faster. I was happy sticking to the lower levels, because for me, it was more about the technical challenge of the project, not necessarily the money.
I would suspect that probably 10-15% of the players on ACR 50nl and below are bots. Most probably run shitty bot profiles, some are probably part of a larger bot ring (these are mostly operated by eastern and northern Europeans) that share hand information during the course of play (i.e. colluding bots). The single-operator bots (mostly US operators) that are good will be good enough in both play and acting 'human-like' to avoid any suspicions. In general WPN and Bovada don't mind bots. As long as the bot isn't the target of multiple complaints from a whole bunch of players, both sites are pretty bot-friendly. Bots pay good, steady rake and seldom complain about anything on the site.
I think I believe him. Enough technical details are provided, the story adds up, and it isn't too outlandish. He's not claiming to have crushed middle and high limits for 6 and 7 figures.
He's admitting that he never went above 100NL, and only cleared 30k.
He is correct that Bovada generally doesn't give a shit about bots. I found this out myself when they were Bodog, I caught a bot with certainty, had pretty much smoking gun proof, and instead Bodog chided ME for exploiting a bug to give myself the button 80 consecutive times against it heads up.
Interesting how this guy adjusted the bot to where it didn't play in predictable patterns, chatted, and didn't bet weird or predictable amounts. Programming the occasional misclick was especially genius.
I can't praise the guy, because he still cheated, but at least he's giving us a look into his process.
Also interesting how he sent the bots to play against lower limit players, fearing that higher limit players would recognize that he was botting. That is also consistent with how a botter would really think.
Recall that Bovada gave a massive refund due to a botting network being caught, where it seemed people recevied a total of about $1 million dollars BACK. I didn't get anything because it appears the bots did not play at 30-60 limit holdem. However, this seemed to occur because it was a massive botting network engaging in collusion, so there were probably enough complaints to where Bovada took action.
More then likely I'm gonna guess he's referring to openholdem bot framework which is well known out there besides the other popular shanky bot software. OpenHoldem is in fact freeware which he's right you have to program intensely. Also everytime the software is updated by the poker rooms they have to reprogram the table maps. I suspect that's why many pay for Warbot which is built on OpenHoldem and keep their table maps updated and just concern themselves with the profiles.. Warbot claims that they have profiles which can actually learn and adjust on the fly even within tournies.. I'm not sure if I believe that but I'm guessing basic AI is within the realm of possibility. I know Shanky bots pretty much are designed for micro limits although they try their damndest to hype the occasional lucksack near or actual bink by their software.. I keep my ear to the ground because honestly the only way to keep ahead of the bots at times is to know what the hell they are up to.Quote:
Originally Posted by Dan Druff
Yeah, I am skeptical that these bots are complex enough to learn.
And on Bovada, they can't learn -- at least not beyond the few hands played against opponents at the same table.
Without being familiar with them, I am guessing that they play fairly ABC poker, with a few modifications so as to not look like bots. This makes them capable of beating donks at micro and low limits, but probably need a lot more work to be able to beat anything tougher than that.
Limit holdem bots are easier to program because the game is more mathematical and easier to calculate decisions, whereas NL often involves decisions made based upon feel.
Bots are never good at reasoning out how aggressive to get with good but non-nut hands, which can be problematic at NL.
There are two signs I've noticed regarding playing a bot (ignoring timing and overly consistent play, which can also indicate it):
- Strange, overly cautious play involving excellent but non-nut hands. For example, bots will sometimes severely underplay non-nut flushes.
- Too many loose river calls, especially at limit holdem.
Bots are also not good at detecting marginal bluffs, so they are often programmed to overly call, thereby making them tough to bluff, and eventually intimidating opponents into not bluffing them anymore.
I used to make some really really thin value bets against LHE bots for this reason.
Bots will get better, much better..
If sites don't pro actively and very PUBICLY go after bots, it will be online pokers downfall.
This is BS from WPN and ACR. My thread was right on. The bots aren't good enough to crush everyone. There's a reason why every time i stole a checked pot I would get dealt a bad beat within a few hands. This is to keep aggressors away from the bots. Also ACR is super card dead. Tiring everyone out and letting the bots handle the work. This bot stuff on ACR and WPN didn't come out until after I made the other thread. Check your old hand history for checked down pots that you stole and see what happened a few hands later..
Just think about it. No need to have people running with super accounts and risk getting busted. We'll just rig the software to be really card dead and let our bots collect for us slowly.. I also remember playing the Super Series... every time I would come over the top of an opponent with an all in raise leaving him with no raise option but fold or call... the board would run out so good for me... I ran top 50 out of 2k people doing this... I also noticed other people doing it and it worked for them also. After i mentioned that it changed.. now it depends on how deep and other ways to trigger it are now in place after i mentioned it publicly.. I kid you not.. I have hand history that will show that...
I dont care how good your bots are.. no way they always draw that good or even win many tournaments..
Rigging + Bots go together...