Two Plus Two Newer Archives - My Report on shortstacking bots NL500-NL2000

Two Plus Two Newer Archives (http://archives1.twoplustwo.com/index.php)

- Internet Gambling (http://archives1.twoplustwo.com/forumdisplay.php?f=30)

- - My Report on shortstacking bots NL500-NL2000 (http://archives1.twoplustwo.com/showthread.php?t=454991)

My Report on shortstacking bots NL500-NL2000

Sorry if I repost this. Will try to do a better presentation so I get some more focus. This is my report about 10bb jam/fold-bots on prima/microgaming No Limit Texas Holdem $2.5/5 up to $10/20. At least 6bots have been identified. They all share the same pattern, gameplay and are from the same city Valencia. They buy in for 10bb, pushes preflop, sits out when they double up.

Here is My report I sent to prima one week ago:

---------

A brief presentation of myself:

24years old
Bachelor degree in Automation and Mechatronical Engineering
Played poker fulltime for 2.5years
Started with Fixed Limit Texas Holdem(played up to $30/60), now playing No Limit Texas Holdem(up to NL $5-10)

Previous experience with bots:
Watched the SNG bots on Pokerstars 20+2. Always sit same seat, push for everything -1 chips.
Was helping with the discovery of bots on FLT 10/20-30/60 on party about 1year ago. 5second delays, 30/17/1.5, 80% defend bigblind. Made some nice money by playing that bot HU on FLT 10/20.
Have watched my friend and his play money bot on Pokerstars.
Have some basic programming background. I wrote the AI for a one-slime-bot for a Java Language Programming Project.

Place:
Microgaming NLT500-NLT2000

Bot names:
deilor
canichewua
piruloo
caird
vbisk
aprilialions

How I discovered them:

1. I noted that deilor and canichewua were short stacking. They were the only short stacking grinders in my games at the time. And the only winning shortstackers.
2. I noted that they were from the same city. My conclusion was that they were friends who used the same strategy.
3. I was playing a table with both of them and got the feeling they were colluding. They never played each other in ~60hands, which is nothing out of the ordinary but a bit unlikely. So I emailed prima(25/6) and told them to do a background check on them. Prima have the staff and access to all the data and get paid for this so I thought they should do the job. But nothing happened for 1week and my 7 email got only standard replies.
"I dont belive the players in question are bots as there are no bots on the network. Microgaming have procedures in place to detect these. Their behaviour also suggests they are human players."
4. Starting to observe them closer and at one time I was stalking them from table to table just to learn how they were playing. I dont like shortstackers and sometimes when I am bored I like to stalk people I dont like.
5. The breakthrough was when I was observing how many players they need for to post the blind for the first time at a table. There were 3people playing in the hand and canichewua was waiting for the blind. When he got to the blind he sat out. So I sat down on the table. Cannichewua then sits in and waits for the blinds. I decided to not to post and sit out and when he was to post the blind he sat out. I then sat in again and instantly, 0.0seconds delay he sits in. I did this a few times just to be sure. A human who follow a strict game play (which before this new discovery was the most probable explanation) would lag a few seconds but a computer with no delay in that particular function wouldn't. Obviously the bot owner could fix this given the information about how they were discovered but every bot's code will have some flaws in it. This was this particulars bot flaw which made it possible to be detected. The other bots I have observed have had even more obvious flaws.
6. At this point I thought it was a very high probability that they were bots so I checked them even further. I learned how to predict their play. I would amuse myself by writing in the chat what the bot would do. "next hand the bot cannichewua will sit out" "pokermanager, there is a bot on my table, next hand the bot canichewua will sit out". I figured I couldnt exploit them without lowering my original winrate had they been deepstacking I might have considered keeping the discovery to myself and tried to exploit them myself. But the bots were lowering my winrate and damaging the fish pool. The botowners, poker manager, and the other players didnt seem to care very much. I got in touch with some other players at the stake who agreed that they were bots and they email prima but no results. Since Prima didn't remove them I tried to scare them away and to post on 2+2, and was crossposted on pokerforum.nu, two major poker forums. The post didnt get much attention probably beacause it was on a non American network. I hope this thread will get some more attention.
7. I noticed that the bot had bugs. For example it once did a full rebuy to $600 and then the next hand pushed preflop first in for $600.
8. I also learned how to exploit the bots tendencies. For example the bot never calls preflop, so I started minraising it. It would still choose between pushing and folding and pushing. Not adapting and sometimes call preflop. Since it pushed less than 20% in raised pot I made auto money on my minraises. I guess this increases the chances that this is not just a very disciplined group of people from Valencia who playes identical, a group of humans would adapt. I got quite good at estimating the ranges and definiatly had a edge against the bot. However the bot did run very good and got lucky many times so in my database I am only +$200 against the bots. But I guesstimate that in skansklybucks I am up ~$500 against it. It is very hard to effectivly exploit a jam/fold-bot while deepstacking a deepstack game.
9. I also got some respect for the bot owner. The bot was very
sophisticated. It would sometimes push a very wide range for example 95o against tight players in the blinds, so I guess it has some Bayesian statistics in its code which I have never seen any other poker bot use before. Maybe the Party FL-bots used it, I am not sure. Or maybe this just was the owner mixing up the bots play.
10. I talked with some other prima players and they added piruloo, caird and vbisk to the list. All of these players are from Valencia and play very similar or identical. 18/7 I added aprilalions to the list, the newest addition to the bot farm.
11. The bot owner seems to know that I like to predicts its play so it has changed the code over the time to make it less detectable. It now will now sometimes play until it doubles up, then sit one extra orbit, then sit out, this one example of its adaptations. I guess soon after this report goes public the botowner will change it and for people who observe the bots will se that the bots play differ from the play in this report. But I have several posters who can verify that my description of the bots play during this period is true. Please do so in this thread if you have played against them. I also think the botowner sometimes play by him self and I guess soon after I post it he will play more by himself. So there will be some problems for people who try to rail the bot which might make my case less believable. Just have this in consideration. Sounds a bit lame when I write this, I know. You should take my report on its own with some salt, but also take the bots action after this report with the same salt and listen to people here who verify my story.

12. Among us bothunters there has been some speculations that __pike__ is the bot owner on the basis that he is from Valencia and plays deep and seems never to play hands against the bots. At the moment we think is unlikely but possible.

My recommendations to Prima:
1. Temporarily freeze all the players in my report so they cannot cash out. And watch them for chipdumping.
2. Analyse your data on them and decide if you agree with my conclusions.
3. Check up every poker player on micrograming from Valencia. Then filter everyone who have played NL100-2000 this year. Then check those who are left for shortstacking tendencies. Examine everyone of them.
4. Add shorstacking to your list of suspect behaviour so you can idenify bots quicker in the future.
5. Increase the minimum buyin from 10big blinds to 20big blinds which is the standard on every other major poker network.

---------------

So why do I write this. I no longer care about if Prima are able to confiscate their money or they will cash out. If prima does confiscate their money no one of us pokerplayers will benefit anyways. I think its more important in the battle to slow down the bots that information is availible so that more bots can be discovered. The botowners will improve their code no matter what, the most important is that the problems gets more attention. Also I want the bots away. The make the games boring and lowers my winrates. Also they are bad for the reputation of poker.

This is my story of me trying to force Prima to remove the bots. I have not been that nice to Prima, in fact I have made some ugly threats about using viral marketing to give them the reputation of being a bot-infested site. But they have ignored me to long so now I make this story public and hopes some action is taken.

EDIT: fixed date typo