All you need to do is detect the frequency of blunders that are outwith a margin of error that could be caused by using different sims. It is really not that hard. The whole point of using an RTA would be reducing these mistakes.

It doesn't really work like that. If you allow the solver to for example bet 25% or 150%, and then run the same spot only allowing 33/66/100, you might end up with two vastly different strategies that are both close in EV.

If the cheater is okay with not crushing LLinusLLove but still making a ton against weaker players they could make a bunch of weird completely different looking strategies that are still decent because the solver is really good at doing the best with the sizings it has to work with.

If someone wants to use RTA to create a balanced flop c-betting range with sizes 25% and 150%, or any other sizes TBH, then I say go ahead, it's a completely pointless thing to do. What I care about is people using RTA to play perfectly in that they are never taking -EV at equilibrium lines, and that is relatively easy to detect.

The point is that the solver can create multiple very different strategies that beat any human. Or do you think that you could beat a solver that uses only two flop sizes?

Why don't people understand this? On flops where everything is mixed a human can randomly pick any strategy and lose no EV to a GTO strategy.

That's obviously not true, if you fold the nuts you lose EV, if you call with the nut-low you lose EV. This is true in a very complex way for all the hands in between too, and even seemingly similar hands are actually different because of how they interact with future community cards, and because of card removal effects.

This wouldn't work against my proposed detection. This would just add complexity to players strategy making it more difficult for non RTA player to implement profitably.

Not sure if you are being deliberately obtuse or what but I am talking about mixed strats at equilibrium. It absolutely is true that you can pick any of these stats at random, because the whole point is they have identical EV at equilibrium.

You could easily build a database of all players with their average EV loss/hand given the bet sizes they use, using only pure mistakes. Then it would be quite obvious who is cheating. Seems very computationally expensive, but any site taking RTA seriously could do this. I wonder if Stars already is.

I am missing how this helps. You have RTA running which gives you an idea at all times what GTO wants to do but you are using it only as a guide. How are you going to look any different to players who have studied GTO extensively? It comes back the same thing unless you are doing a Mike Postle impersonation online you can easily fly under the radar.

Still think some version of Poker in VR is one answer. It doesn't have to be in VR but something that tracks everything similar to a VR headset. A VR headset like the Oculus Quest is tracking EVERYTHING. It can track and map the entire room you are in, it is tracking eye movement, body movement, everything done on the device, and has cameras recording everything.

No.
I don't know what it is with you guys but these threads always get cluttered with wrong information about theory.
Please stop. It's tilting to read all of this.

anything short of all decently winning/high volume regs playing with a 360 degree camera can easily be sidestepped.

the database backtesting would be interesting, they're able to do it realtime in chess. The poster talking about just using non std sizing schemes is right tho i think, ive always assumed most poker rooms look at online poker as a managed decline and not smt to invest in too much into anyway

Yes, the number of people who don't realise all mixed strategies have the same EV at equilibrium is staggering.

That's not what you said in the post I quoted.

Sorry, I find no pleasure in correcting the flaws in your logic. Let's just leave it here.

The context of my post was playing against a GTO opponent, which is obvious if you read closely the prior posts. So yes that was what I was saying.

This is actually not true really at all, a solver strategy with incorrect sizes will still have an EV loss which is way below elite human levels vs the equillibrium strategy.

Sites should (and many surely already do), not look at the frequency of actions or sizes chosen but the EV loss of actions vs equillibrium.

Can even use a small subset of the gametree (3bps) to get a good idea if someone is cheating I think. There are many other low filters unmentioned in the thread to flag potential bots, but outside of Stars I'm not sure many other sites have caught on and even so the barrier for legal balance confiscation (evidence that holds up in court) is much higher than various red flags being hit I think.

Isn't this baked into the contract that you make with the site that they have the right to seize funds at their discretion? Even if the law supersedes this the cost and time for a player to fight this in court against one of the big sites doesn't seem very likely proven by GG's willingness to just snap confiscate?

I am going to say it again as the more I think about this the more it makes sense. I think even someone could take on the project. A VR headset is tracking everything and Oculus has got the cost down to $300. Take all the VR stuff out of it and just make a headset that tracks everything. You can probably get the cost to less than half that amount.

It would have so many applications in gaming, chess, poker, and whatever other online stuff is done competitively that it would have a very big market. The poker sites could give them away free to players. You could introduce games for players who specifically use them and while people make the point that recs don't want such nonsense then let's see how many players flock to games they know are clean and are being highly monitored.

You are obviously going to have some logistical issues with the device being hacked but I'm pretty sure you can make some very good safeguards against that if done properly.

Sites should invite winners to take quizes with questions based on their past hands. If your answers don’t match the play you made recently, then there is a problem.

I think the sites do something like this occasionally in that if they have someone they really suspect of RTA then they will ask them to play some games on a stream monitored to see if past play matches up. I think though this is extremely rare circumstances. Of course if your GG you just ban them no questions asked.

The solution needs to be better than reactively addressing accounts as they get flagged. Players who are going to these lengths to cheat have no problem finding new accounts to open and you are always playing this game of cat and mouse. It probably solves problems at the highest stakes where the player pools are smaller but the games are going to become invected at all levels if they are not already. You need a real-time solution imo.

I like this idea... I hardly play online poker anymore, because I am concerned that I am playing against bots and/or people using RTA. I would definitely invest in a headset if a site adopted this concept.

Are you suggesting sites to catch cheating before it happens? Alerts/processes/investigations are... already in "real-time". As for brand new accounts- it's really not easy to multi-account.

I would also say highstakes games are almost impossible for any site to police, unless the cheater gets carried away. Main problem (and it's getting worse) as always- sample size.

HU in particular- in case anyone is reading- please don't expect sites to spend tens of thousands solving for particular trees for somewhat arbitrary stacksizes for specific match ups. Ending up with sample size of 2 at best in various nodes. I def recommend playing HU on Stars vs anywhere else, but there is no security beyond a "standard level".

Good thing at least that if someone is taking it too far (high volume at hs, a ring of cheaters at hs or using seating scripts!!) they will get caught for sure. Single table RTAs isn't that high EV, and unless you are a hermit, someone will rat you out eventually

It's actually the anon games, followed by random seating (zoom/spins) that are safest on Stars. Note- safest of course doesn't mean highest EV to play.

I don't think we are quite there with the 'Precrime' system just yet

I am talking about systems that provide irrefutable proof. At the moment you have retroactive reviews, monitoring, and maybe stuff like session streaming to show the validity of play among I am sure a whole host of other things. They are doing nothing against the current threat of a seasoned poker player having RTA running as a guide but using it intelligently to cheat. NOTHING. I have yet to hear one argument that shows how these guys are getting caught.

RTA guys are getting caught. I worked on such cases when I was at stars.

It's just difficult, and results are never communicated. And as I said being realistic, enforcing sporadic RTA look up using various libraries of solutions public or private is impossible but risk-reward is not so obviously tilted in favour of RTA users yet. Maybe you misspoke when talking about catching them in "real-time".

If you expect sporadic RTA lookup to be caught by sites with irrefutable proof- this is just fantasy. It's by definition something that doesn't have enough sample to mean much, and doesn't gain a huge amount of EV. The more you mix sporadically the longer it takes for samples to converge, if at all.

In terms of the actual methodologies of catching bots(aka RTA), you won't find much publicly here in this thread (at least any useful idea will be amongst piles of other posts).

I am not talking about sporadic lookups. I don't know why when we talk about this topic it seems to be really hard to get straight-up answers and people seem happy to obfuscate the questions and responses.

Let me be clear I am talking about a player who is screen scaping the cards in some way in another program and has real-time assistance running showing him the solver action in each of the spots he is playing. He is a good poker player who is doing a fair amount of deviating maybe for exploitative reasons, he doesn't feel a human finds that line very often or whatever. He has the RTA mainly running so he knows always in tough spots where he is in his range etc and how each spot roughly plays. This person is impossible to catch with current methods.

There are no current methods that allow this to get detected. It is literally impossible if the player is clever and has the technology setup correctly.

Edit - I outlined in a previous post how using something like the headset system would provide that irrefutable proof and obviously stop it completely in 99% of cases. Whether it is viable or not I don't know but the point is that a completely new approach of some kind needs to be taken to the problem.