While no superusers have been found (on sites of any reasonable size) for approaching a decade now, that is obviously still a topic that a small portion (typically the riggies) bring up as a potential boogeyman in the industry to go along with their standard action hands, new player boomswitch etc. nonsense. A recent riggie, TheoryJuicer, refuses to post his math questions in actual math forums (Mike Haven had to move one here for him), so since he just asked us to post for him, I thought that was reasonable request - here it is.

He maintains that superusers, if played properly, would never be caught, and the proof essentially is that no superusers have been caught in the past 10 years. Yeah, this is a bit of a "I cannot see an invisible person, thus it proves they exist" form of logic, but looking past the standard riggie logic issues, he eventually suggested that the perfect superuser would be one that wins at 5 bb/100 hands so as to never get noticed.

Mike Haven said that type of user would be caught, as I did as well, though allow me to clarify my reasons (Mike Haven can present his if he likes), and then I will ask the math guys here if they believe that a super user winning consistently at 5 BB/100 can be caught.

Note, the debate also involved some programming concerns by TheoryJuicer, and my suggestions that he post those in the programming forum were of course ignored, so if any have expertise on that here, feel free to stretch those muscles as well.

My beliefs include the following:

- The superuser would need to play a buy in level game to make it at least worthwhile as a form of income. Playing 25NL would make zero sense, so I assume the 200NL+ games and above would be at least worth the effort as 5BB/100 on the Zoom games could be about $10-15 per hour per table. 200NL is probably still a bit low, but let's use it as a baseline.

- The player pool at those levels feature a lot of players who do a lot of hand history research, and as well they are keenly aware when a new player becomes a reg and will look at their large database of hands (and ask other players) of any history on this new player to see if they moved up the ranks or not.

- A super user will have to play differently to win at a 5BB/100 rate than a normal good reg. They will win hands in ways that are different than normal play (where one cannot see all the cards). Thus, my belief is while some overall stats can be created (VPIP, PFR for all hands type stuff) to look like a "tight aggressive" player, when one digs deeper then vast differerences in how a player got there (position based stats, post flop play approaches) will stand out very noticeably and quickly.


Note, TheoryJuicer said he would post stats of people in his database with a win rate of 5 BB/100 hands to prove his point. He of course never did that, but I suspect that different win rates are possible depending on the buy in level of the game, so if he plays 2NL or 5NL as his posting history seems to indicate - it would not shock me if 5 BB/100+ players existed (human and bots), though I have a hard time believing any would be superusers at those levels.

Thanks in advance, and I hope this question makes sense, as I am curious to those with a much healthier foundation in math provide their take on how easy it would be to catch a superuser trying to pretend to be a typical winning "tight aggressive" reg in the mid to higher buy-in cash games.

I have no idea of whether or not this type of user could be easily identified. I just find it hard to believe that a superuser could play at a consistent 5BB per 100 rate. They would have to intentionally lose a lot of hands just to maintain that low of a winrate. Excluding rake, a completely random player playing against random players will be at 0. All you need to make 5 BB per 100 is to win 1 small pot out of every 100 as a result of knowing your opponents' hands.

I think it would be much more of a struggle to win that little (on average), given the extra knowledge, than it would be to just play good poker and take your chances. If you paid no attention to their cards on 99/100 hands, and LOST 10BB, you would only need to win 15BB the one time in 100 you decided to use your advantage. Can't imagine a superuser playing that way.

Yeah, that is the part that I have a hard time believing as well. The proper superuser approach is to smash and grab as they did a decade ago. I do not think that can be as easy these days (money does not move as carelessly in such quantities), but I just cannot see how someone who sees all the cards can mimic someone who does not see all the cards. Even if they somehow limited their win rate (which is silly, as this is not a long term criminal approach) - it would show in the data of how they play their hands.

A human superuser winning so little is very hard to imagine in my opinion. However a superuser bot would be very easy to program and next to impossible to detect. You just tell the bot to make its usual game pretty much always and peek cards some (small) percentage of the times at the river. This could be enough to have a consistent and undetectable winning rate.

EDIT

When I say next to impossible to detect, I mean to detect as a superuser of course. It could be detected as a bot from other reasons of course.

From a practical standpoint, I think a smart superuser would be switching accounts after every big tourney win. Running like the sun in one tourney isn't going to raise any flags. If I were looking for superusers, I would look at accounts that popped up, quickly had a big score, and then never played meaningful volume again. But it's impossible to say how often that legitimately happens, so I don't know how you would ever figure it out as a random dataminer.

If the premise is that some buttlicker is afraid of superusers in his cash games, basically the same protocol applies: new account, run like the sun for 800 hands, maybe give a little back so it looks like your hot streak ended, cash out, never play again. I assume there are people who legitimately do that, so a superuser-hunter could never be sure even if he found recurring instances of that happening at his tables.

I suspect that behavior would be caught for a different reason (the extreme results on a sequence of accounts), but the premise here is that the superuser would blend in for a long period of time, winning at 5BB/100 or so for weeks, months, years etc without being caught.

Your suggestions seem to imply the above would not be possible (ie: the winning at 5 BB/100 thing) without being caught, but you can choose to explain more on that if you prefer, but even the way you are describing how it should be played would stand out quite a bit.