The detailed stat comparison screenshot shows that stats of different players from the group, in particular, SB/BB VPIP, flop cbet/raise %s, are often much farther apart than the length of confidence intervals.

I assume botters are too dumb to make artificial delays in reprogramming of their bots; when they enhance the algorithm, they implement the update for all their bots at the same time.

Perhaps it means that some of these accounts were playing 2-5-max much more often than the other. Why?

Or do you suspect that there's a pattern of relative seating of these accounts, i.e. one of them is ordered to sit always to the left or always to the right of another account? Or do they take seats randomly?

Calling into question the value of such a bot creators work is not to say that no one would do it and does not also question whether or not this evidence is good. Your original question does not still stand, calm down.

Why you assume botters are dumb?

We haven't analyzed their seating behavior.

I will concede that I may have slightly misinterpreted cbt's original post and also that my posts may seem a little aggressive towards cbt after the fact which was not my intention. I agree with cbt that if our hypothesis is true, it is a somewhat impressive feat.

try to chat with them..

It doesn't seem far fetched that somebody capable of programming a winning PLO bot would also be capable of taking precautionary measures against the detection of said bots.

The accounts in question do not play the majority of their hands at the same tables (to my knowledge). I believe if anything, that the variation in stats could best be explained by playing different sets of opponents and in different positions relative to their opponents (maybe you meant this).

Why do 5 accounts with near identical stats (namely VPIP/PFR/3B/WTSD) an indicator of automated play?

You can find another 5 accounts with near identical stats (ie 30/20/12) and claim another bot ring.

Can you show more detail stats like positional PFR or flop stats? Otherwise there could be some selection bias.

+1. Coon74 explain it very well

As most of the edge comes from sharing hole cards, that they avoid seating with each other too often means that they're not so dangerous anyway, doesn't it?

Also, I think such radical precaution measures would be overkill against iPoker's security department Creating a ton of accounts and seating them in an unpredictable way would suffice, but never occupying a juicy vacant seat for so many minutes in a row is not smth an automated collusion ring would do imo.

they chat very normal with me.

A bot could click on hundreds of tables in less than a second. It wouldn't matter if the windows are stacked/hidden.

edit: I think I misunderstood what you wanted to say.

It's plausible that they are table selecting normally, but colluding when they happen to play together. Also, identical preflop stats can definitely be played off a chart, but the farther you get into postflop the less likely that is. No two humans (let alone 5) could, imo, have the same stats on all 4 streets.

Are you really stating botting is acceptable as long as you don't collude?

We posted more stats than just VPIP/PFR/3B/WTSD.

There are no 5 accounts with >=30 VPIP with similarities like these accounts in the data. We have a control group filter for that. Usually if you find an account that is almost identical, it is just another account of an existing player. But these players have all played against each other and are therefore unique accounts.

The main argument is not collusion, it is botting. I believe these players have an edge pre collusion. Collusion is speculation based on the information at hand. If somebody is willing to violate the ToS and in doing so deceive both the poker host and his opponents, you have to assume they are also capable of colluding given the opportunity to do so.

Furthermore, a bot is still problematic to face as it won't give up certain edges that humans do such as fatigue, misclicks and tilt and will therefore have an unfair advantage over human opponents in that regard.

I agree that the accounts in question are clearly not maximizing their EV, at least not in the short term...

There are plenty of pre flop and post flop stats on the opening post.

You are free to show us five PLO regs whose stats are as uniform as here. Even two is ok for a start...

I see 3 of the 5 listed on Titan Poker Rake Leaderboard for this week.

http://poker.titanbet.co.uk/results/rake-race.html

Players of this caliber and volume are usually members of online poker communities. I wonder if they have been made aware of this thread and if so revealed their 2+2 usernames (or other online account) to defend themselves?

Also to the skeptics, what would constitute "evidence" in your mind for 1) bots and 2) collusion?

Well similarly to the other thread, freewilly and Isaac Asimov seem to be the only 2 people convinced of this.

I am sorry I missed the post flop tendencies. I am convinced there is some type of foul play now.

these are my stats for 3 of the 5 players, 2 of them i haven't played much at all

the screenshots are from a week or so ago, cant remember the order of them but they are fed, stone and anarki

Thanks, napsus.

I'm a bit of a novice when it comes to interpreting the stats, but I see a striking across-the-board similarity between the three accounts, particularly when comparing how they play at the various seats.

Do the figures also come fairly close in very small sample sizes (say 200-500 hands) as well? Or are these bots just scripted to play a specific range of hands that will in the long run tend toward the stats shown? I ask because I'm wondering if it's just a bunch of people in a grindhouse all directed to play with a narrow set of parameters, thus creating remarkably similar ranges for all seats and common situations.

For those wondering why these bots aren't better served at higher stakes, is it reasonable to conclude that what appears to be a TAG-ish, more straightforward range is more effective at this particular stake? And that these bots would actually get exploited at bigger games? Or could they be tweaked to adjust to, say, 500PLO?

Anyway, sorry for the noob-ish questions. This stuff fascinates me, especially since I'm now more than three years removed from online poker.


Edit: oops, one more thing. That video is meant to provide evidence of similar timing. With the unaided eye, I don't really see much on there that constitutes any evidence (other than the "damn, they're all slowpokes" observation). If these are truly bots, you should be able to compile a dozen different videos of the suspect accounts, synchronize the starts, then play each at real-time speed to show that they take exactly the same amount of time to raise, or 3bet, or fold, or whatever you're demonstrating.

Spoiler alert for those who haven't seen The Hunt For Red October:

you are definitely wrong

Let me get this straight, in the initial thread in Internet Poker one of the main arguments freewilly made for there being something suspicious going on was that they ran way over EV. So now I presume you guys are basing your suspicions off of their similar stats rather than their rungood (?), since obviously botting doesn't explain a group of players running over expectation.

I'm just pointing out that your arguments are not consistent, at least not across both threads. Are you still holding to the suspicion that this is a hole-card sharing botnet or not?

We never intended represent the timing tells as strong piece of evidence.
Similar timings from a group of players with matching stats was the thing that got me analyze them in the first place.

I am sorry but we are not hunting a nuclear submarine here so our resources are limited.
If you are willing to compile some videos it would be a great help though

Any fool who plays online poker now deserves what he gets. Bots are only going to get better and more difficult to detect, but I've lost track of how many other ways people have been scammed and cheated online, not to mention simply being unable to cash out, and that's just from casually reading 2+2.

The other thread emphasized my findings of EV disparity over large sample size and lead to the fact that we found five accounts with matching stats.

The stat similarity itself is a strong proof of automated play i.e botting.

We don't rule out collusion but we can't prove it either with this sample size.

Can you be more specific where our arguments are inconsistent?
Do you mean it has to be either a colluding bot ring or no bot ring at all?