Poker-AI.org

Hello there,

we are the team behind https://rta.poker, a new real time assistant for online poker that we've been developing for quite some time now.

It's probally best if you see it in action yourself!

https://www.youtube.com/watch?v=ZaFseiEBVAs

Now let us explain a few things about the product.

Lets start with probally the most important aspect of all:

Stealth/Security

This is by far the most important feature, after all no one wants their account restricted.
We've made sure to do everything possible to be undetectable. EVERYTHING.
We are running a dual computer setup.
A computer on which only the poker client runs, aswell as a second computer to run the RTA itself.
The only connection between both machines is using a video capture card that is vurtually undetectable from the poker client PC.
The capture card will clone the EDID (Extended Display Identification Data) of the display connected to the poker pc, so the poker software doesn't know a capture card is connected.
Doing all of this leads to the software being undetectable.

Solves

Without us having the information of who did what out there, it's tough to assume where we stand in terms of solve quality.
We do however think that we've created one of the best preflop solves anyone has ever put together! The GGPoker preflop tree we've built has all the bells and whistles you'd expect (preflop rake included, 1k NL rake structure)
from a top quality preflop solve. We have calls in ALL spots (which is kind of a big deal for a preflop rake environment like GGPoker, calling becomes a lot more
frequent due to the disadvantage preflop rake introduces compared to normal rake structure games (e.g. Stars/ACR). This also means that we're not mapping calls to different positions
(example: UTG opens -> CO calls -> everyone behind folds, most services will either lack solutions for this spot or map CO's range to BTN because they don't introduce CO call vs UTG
open in their preflop trees (as this requires huge amount of server resources/time to solve)). Without giving away too much information, we'll just share that the preflop tree required
close to 1TB of RAM and over a month to solve to 100 iterations/node (recommended I/N by the developer of the software used to solve the preflop tree).

You can expect the same high quality from the postflop solves as well.

SRP: solved to 0.5% exploitability
3B: solved to 0.3% exploitability
4B: solved to 0.2% exploitability

Keep in mind that we're resolving each spot from turn onwards on the fly to the exact game scenario you are facing on the table (we solve for exact betsize villain bets on the turn, not mapping to the closest one we have on hand) for the current hand on our robust server infrastructure.
This also means that the exploitability we just quoted, goes even lower!
Another feature that we offer is custom turn/river sizes for each customer.
An example of this would be Player A uses the default structure and always has 33, 50, 75 and 100 percent bets on the turn, Player B can use the same default sizes or decide to go with 33, 66, 100, 150 (or any other they choose!).
Anyone looking over their games under a microscope would have a hard time identifying them as "playing similarly" or "using the same tool". This improves stealth a lot!

Performance

The best option to judge our performance is by looking at our demo video.
Keep in mind for postflop solutions we are resolving on the fly, and we spend countless hours improving our server infrastructure and algorithms to provide solutions in a matter of 1-2 seconds. We are certain no one else on the market comes even close to this, providing the same detailed solves we do (if one goes lower amount of sizes, or exploitability obviously they can provide faster times, but not better overall quality).

UI

Not a lot to say about this one. We tried to keep things minimalistic and rendering your poker client PC display onto your RTA PC.
This has the added benefit of playing on just one monitor (even though 2 PCs are used) and to be able to easily react to captchas/gameplay/chat messages/emotes or the likes.
We do recommend going for a higher end capture card if you want to be playing on just one monitor though, as most cheap capture cards have some input lag which might be annoying to some people.

Hopefully this is enough to get a decent idea of what to expect from our product. Now let's do a quick FAQ.

FAQ

1. What poker sites are currently supported?

-- Currently we support GGPoker aswell as the WPN network on ACR including it's corresponding skins like Pokerking and Blackchippoker. We are also working on adding most of the big networks to our tool in the future.

2. How many tables does the RTA support?

-- We are now supporting 6 tables simultaneously.

3. What do I need to be able to use the RTA?

-- To be able to run the tool efficiently you will need following hardware:
1. A PC to run the poker client on, for this pretty much anything works aslong as it's able to run the client with a stable 60 FPS.
2. A second PC to run the RTA on, for this we recommend following hardware:
- 8GB RAM
- Nvidia GPU with atleast 4GB VRAM, currently AMD Cards aren't supported.
- A fairly modern CPU. Anything that came out in the last 6 years should be fine.
3. A video capture card, preferable with an EDID cloning function, a low end card will be sufficient, but can introduce some input lag which for some people might be an inconvience, for those we recommend a higher end card (eg. elgato).

4. What happens if my opponent bets some really strange sizes preflop?

-- Usually we map preflop actions to the closest preflop sizes we have solved for, if its too far off, for example UTG opening 5BB we have some "OUT OF BOUNDS" restrictions implemented so you do not get shown a faulty solution.
Keep in mind that against those opponents it will most likely be the best option to play exploitative without the RTA regardless, since they are most likely fish.

5. Does the RTA support 3Way+ action postflop?

-- Not at the moment. We do have plans of introducing such solves, however they don't make as much sense as they don't guarantee profits in the same way a HU equilibrium does, so it is low priority for us.

6. Does the RTA support limp pots?

-- No it does not.

7. Im a new poker player, I struggle to beat NL2 online, will I get rich quick and beat everyone if I use the RTA?

-- No. A RTA won't make everyone a winning player, regardless of how good the solves are. It is a very strong tool, which if used correctly, can lead to substantially more income, but it's not magic.

8. I am a bit sceptical, do you offer a demo version ?

-- Yes. We are offering a 2 day demo version for everyone who is interested in our product.

Contact

[email protected]
https://discord.gg/pd6kzq5tpz || crownmaker#3948

Hi regarding stealth, is user required to set specific table layout, table size etc in order for OCR to work?
What does it make "the most advanced" over competition?

Hello and thank you for your interest!
Our tool does not require a specific table layout or table size (no overlapping though so OCR can work). Generally speaking bigger table size should equate to lower count of OCR errors, but it's working very well even with the smallest table size GG allows.
We do require a specific card design aswell as 4 coloured deck to be enabled, we're working on adding more card designs in the near future. This would further improve stealth and as you know that's the most important thing for us.



To be totally honest this was more of a marketing/slogan thing. We don't claim to know every other RTA product out there (most are private).
Some of the features we've implemented though we have never seen outside of our product (example would be custom turn/river resolve sizes for each of our customers which you can change in the settings).
Another example highlighting the quality of our product is that we're not secretive about our stealth/security features since we're that sure even if poker sites know what we're doing, there is just no way to detect/do anything about it.

If you have a feature that you think is good for either EV or stealth, don't hesitate to request it and we'll do everything we can to implement it for you.

Thanks for the answers.
Did you consider automating the play when you have the solution and letting user play when you dont?
Would that lead to ban?

Considering that you do OCR, do you think the issue could be if poker site client would change positions of the cards etc or or change design during the game, and then see who is quitting the game, or suddenly diviating from standard play?

Regarding your opinion that RTA would be undetectable even if poker site sees all your measures. I think they might not detect you by detecting RTA executable / activity, but maybe due to comparing player stats in HU spots with other pool of players that use solver, and see how often your play deviates from solver suggestions. Player who follows strictly solver suggestions in HU spots would generate after e.g. 20k hands played very similar stats to other players who use solver. It would be similar approach as detecting bot rings using player stats.

This has a few problems/challenges associated with it.
First, from a technical standpoint it introduces a few challenges. For example it would be tough to make the mouse move humanlike. While there are some good algos out there for approximating this it introduces other vectors for GG (or other pokersites as we'll introduce new ones soonish) to possibly identify foul play.
Second, this is very likely going to reduce the EV of the player for the same number of hands played. GTO solves should pretty much never be used vs a weaker opponent as it should be very easy for anyone playing poker professionally to 2x-4x the EV vs that player compared to the GTO strategy. Being balanced makes it mathematically impossible to have negative EV longterm vs anyone, but also makes it that you have a 'ceiling' of what you can win vs bad players. This 'ceiling' is quite a bit lower than one might imagine in a 6 max game. Even vs regs there are so many obvious spots where people are just not bluffing enough (you tend to notice those quickly with experience) hence we can just not bluffcatch marginal calls (+0.00 EV spots, GTO does a lof of bluffcatching in certain spots) and pocket the EV we just gained by not listening to the RTA. Of course one can say that you'll gain this EV back by just letting the RTA (well, a pure bot at this point, shouldn't be classified as RTA anymore) play for a lot longer but this has issues in itself which I'll touch on a bit further down this thread (talking about the statistical detection approach).
Third and probably most important it totally forfits our OCR/2nd PC model that we've chosen for best security (you will need something installed on your GG client PC to be able to control it, hence removing the purpose of the dual PC setup).
So it kind of lowers stealth/security by a huge margin while lowering the EV of players (at least for the same number of hands). We don't see that as a great addition to our tool.



Never thought about this, seems rather unlikely to ever happen. Honestly, if something like this happened to me as a normal user, there is a high likelyhood that I'll probably stop playing for at least a bit while figuring out what just happened and how to convert back to the cards I'm used to.



While technically possible there is nothing (very little actually, which we've already done) we can do about this (by extension any other competing RTA product can do about this).
The only way to prevent statistical detection approaches would be to either not play at all (so no stats for them to compare, but this obviously isn't a solution) or fastly diviate from what GTO is doing/wants to do (again not really an option if you want to gain EV in any way).
However there are a few caveats. Hands required would be far far more than the example you gave if we're talking about detecting postflop strategies. For preflop similarities 20k hands may be enough, but GG would light up as a candle if that's the case since basically all regs would use very roughly the same preflop ranges for all spots. For postflop (flop in particular, as we said we've implemented turn/river custom sizes for users and they can actually do stuff to hide their similarities to other RTA users on those streets) 20k hands means you'll get to play roughly 200 hands (yes it's around 1% of total hands, you can check this in your database) BTN vs BB in single raised pot where we'll most likely have 1 hand played on 200 different flops out of 1755 strategically different flops. Maybe some flops have 2-3 hands. Still nowhere nearly enough to make any conclusions with any certainty I'd argue. There can be more complex approaches with bunching/bucketing different spots/flops together but it's getting very complex very fast and seeing how it's not guaranteed that you'll be close to the database frequencies for any RTA project in any particular spot anyway... probably not the beast approach.

This last part was trying to persuade you that it's neither as easy nor as fast for someone to be detected (unless you run the software literally on your GG client PC, then yes, you CAN get banned very quickly but seeing how some of the competition is still around even that's probably not happening with the speed of light). Fact of the matter is that we've been running extensive internal testing before building our website and offering on this platform. Not a single user has been banned/warned/reviewed.
Can we guarantee that this will never happen? Of course not! Anyone offering you an RTA and saying they can guarantee it is lying as you can't externally beat statistical approaches. But if you're worried about those this (or any other RTA product) is likely not something that should interest you. As everything in life you'll have to choose if the potential gains outweight the risks. We do belive we're offering quite a good deal in that regard.

Regarding detecting players by stats, bots of my team are based on live solver and apply exploits using player stats on the top. Even though different accounts play vs different human players and this way apply different exploits both preflop and postflop, and using different bet sizes on turn and river when live solving, when observing most common stats e.g. cbet flop, turn, raise, flop turn etc those stats are quite similar even after 10-20k hands. I agree that only by looking at preflop stats, it is not wise ot ban accounts as most regs follow similar charts blindly without much deviations, but looking together at preflop, flop, turn stats and seeing that e.g. 50 stats differ only by few % total is already strong indication. I have imported hand histories from around 15 of our accounts playing on the same poker sites, and displaying in PT4 around 20 most common preflop and postflop stats. was surprised how easy it is to detect similary of accounts by looking at the stats even as human using such report, obviously machine learning models would detect such similarities much better.
When comparing 2 human players that play without any RTA, you will see in multiple stats big differences in tendencies e.g. one check raising agro on flop while other very passive, one folds a lot vs probe bet other doesn't etc while when comparing solver players or bots you will not see such big differences in tendencies, even if you use totally different bet sizes for different players you will never have 1 player check raise super passive while other very agro.

We can't really comment on what you've build and why it's getting detected (we presume it is, since you are mostly trying to put forward the idea that it's way easier to detect someone than it truely is) as we don't really know what you've done. Just as a quick comment though anything bot related (read automated play, unlike RTA that just give suggestions) as you seem to have built is going to be vastly inferior to our approach (2nd PC + OCR), where we don't have to have software installed on the poker client PC.

Following an algorithm 100% of the time is certainly not the best approach to avoid detection. What we offer and the way we intend customers to use it you won't be following the RTA in more than 50%-60% of hands, a low enough number that real player differences will start popping up in the stats. So comparing the detection time for one method (BOTs) vs the other (RTAs) doesn't really make sense. If you got detected this fast, it's very many times more likely it's due to the way your whole setup works (like running some software, however small, to control the poker client on poker client PC) than because you did 20-30-50 or even 100k hands and got caught by some statistical approach. (again saying this without knowing what you actually did in reality, might be something simple you did wrong and did actually got cought becase of a statistical approach, or using VMs, or same hardware/IPs as last time, it really can be a million things, it's just that the 20k hands you did is the least likely one from where we're standing)

I didn't say that our bots get banned. We run bots on the sites that want to increase traffic or on sites with low securty. Obviously it is much harder to be undetected on high security sites using fully automatic bot comparing to RTA. My point was that hiding restricted software from poker client is fine, but that poker site could detect RTA using player stats analyses, regardless if using fully automatic bot or RTA. One thing is running few users on one poker site if you just started with your product, other is running e.g. 50-100 users as then those 50 users would get over time more similar stats. It could be off course that GG poker is not doing any stats analyses of players, from what i know GG poker has lot of bots. Pokerstars and party poker should be a better test for your stealth.

Thank you for your clarification, we do intend to offer both Stars/Party in the next 2-3 months (almost ready to start internal testing) and if testing goes well and we feel we're dealivering a good/safe product, we'll be sure to share it here and on our website.

According to multiple reports a few years back Stars bot detection changed the card images slightly during play. If your bot couldn't cope and stopped playing you'd raise a flag.

My bot is saving each table screenshot when doing OCR, and on party poker few years ago, before being banned, i saw a table changed font size of bets / stacks and also design of cards all of the sudden. My bot timed out and closed table, which lead to quick ban.

It is safer to sniff traffic that game server sends to client, instead of OCR.

Thanks for clearing that up,
We use our own trained AI model to read the values of the cards, instead of, for example, some sort of template matching. Obviously if we can't test it we can't guarantee if it will work for a "slight" switch (depends on what "slight" is exactly, but we can't give guarantees). But there is a high probability that it will just continue working without a hitch.



Your example presumes once OCR fails (which granted, it does from time to time for various reasons that I won't get into right now) you're just caught/banned which is far from reality as we always have a human behind the weel, after all they can easily just continue playing for 5-10 minutes on his/her own and close the tables. Remember we don't have anything installed on the poker client PC, you (a human) always have to input RTA advice/his own decisions. If you're trying to build a stable of bad poker players that can't hold their own vs the pool then this might be an issue for you (if/when it happens on certain sites) as you'll lose some EV in the 5/10 minutes where your horses will continue playing without RTA advice. This is generally a bad example anyway as even if RTA works 100% of the time these players will probably lose enough EV in multiway/limped pots to be an unprofitable venture. So it's kind of a given that the player behind the wheel needs to be a player that is somewhat comfortable playing the stakes he's playing.

For our target customers though, people that are already playing mid/high stakes and just want something like a "HUD" that just tells you what's the GTO strategy in that particular spot is (and an RNG added to that if you want to follow it blindly) they can easily just continue playing for a few more minutes/half an hour and can then stop and inform us what happened. Once we get a word/more information about an issue like this we'll implement an improvement/solution swiftly.

This product is not a bot, it's an aid to your everyday grind. It will improve your results if used wisely/correctly. Regardless of how it fails you can still just continue playing on your own on the poker client PC (even if RTA crashes or can't read anything due to pokersite security changing fonts/cards whatever).

I was interested in an RTA but then checked out your website and was disappointed to see that the only results you show are two low sample graphs where rake is not included. After including the high GG rake they are breakeven graphs.

If these two low sample graphs are so good that you include them first thing on your website, that proves your RTA is not working. You're basically saying if I pay you every month and risk my account and my funds by using your RTA, I will be breaking even. If that isn't a scam then what is.

I guess the reason you show these two graphs is because you have no actual customers that win and/or stay for a long time. You almost got me but I am glad that I visited your website.

Yes i agree having human doing the play using RTA is what overcomes those issues with replacing design of table. As the user is not using any bot, but rather just additional software as hint. The only way to detect user using RTA i think would be to compare
each play with what solver would play in such spot and conclude how often the strategy deviates from solver suggestions.

Thanks for taking the time to register a brand new account to shed light on our "scam".

I'll still answer your concerns directly. First and foremost we purposefully never advertised anything related to profit and/or winnings. We even have an answer in the FAQ section specifically for people like you that don't understand what we're offering, I'll just copy/paste it below:

Q: I'm a new poker player, I struggle to beat NL2 online, will I get rich quick and beat everyone if I use the RTA?

- No. A RTA won't make everyone a winning player, regardless of how good the solves are. It is a very strong tool, which if used in the right hands, can lead to substantially more income, but it's not magic.

Everyone knows GG has high rake (including you it seems, since you said it yourself). There are a number of threads on poker forums (won't link but google is your friend) that show trough datamining websites that GG is (compared to other networks) not beatable for midstakes+ (excluding nosebleeds we guess as rake does go to under 1 bb/100 at some point) before rakeback/leaderboards/other promos. The vast majority of top volume grinders consistently lose prior to additional benefits (again, rakeback/leaderboards etc.). This of course doesn't mean that GG is not profitable to play. "Breaking even" at 200 r&c easily means close to $6-$8k EV/month for a dedicated grinder (depending on how much you manage to optimize leaderboard income, 100-120k hands/month).

About the graphs... We (and everyone else sellings RTAs/courses or wahtever) can easily fake the best graphs in the world if we were looking to scam someone (which considering the mediocre graphs we're showing kind of speaks volumes contrary to the "fact" that we're trying to scam someone don't you think? ). These particular graphs probably won't ever be updated/improved as they were from our internal tests, quite a while ago at that (we just needed something to fill the space on the website, and they happened to stay there, for now at least). We're just the type of people/team that likes to underpromise but overdeliver. We would also never ask a user to send us his pretty graphs so we can upload for others to see, this goes against everything we're trying to build/achieve (has implications on security at the minimum). Also, we fully understand that none of our customers have any interest to advertise our product (this means more RTA regs on their tables, hence less EV) so we won't ask our clients to come here and spam (this is our job (lol)) how successful they are with our product, you won't see that here. Just as a general advice though probably don't go buying stuff just because someone is showing pretty graphs, try their product (we do have a free trial!), understand what it's doing and if you enjoy it and see potential EV, try it for a full month, if you don't see/understand where this might be of assistance, just skip it and buy that course from this coach that won 400k in 35k hands (with graph proof of course!)



Yeah, there is pretty much nothing that can be done about this. Every RTA/BOT or even a guy randomly opening GTOwizard/PioSolver on the side will have this potential vecotr to worry about. Heck even super good regs that's trying to mimic GTO plays (and does so succesfully at some decent frequency) will have this particular issue.

This is actually good to hear. I was worried about the future of online poker but if your customers pay thousands of dollars to break even then I probably don't have to worry so much about playing against them

While the tool currently supports up to 4 tables, rta.poker is actively exploring the possibility of increasing this to 6 tables, catering to the needs of avid multi-tablers. For those who remain skeptical, rta.poker offers a 2-day demo version, granting players the opportunity to experience the tool firsthand and witness its potential to enhance their poker gameplay. As poker enthusiasts continue their quest for improved strategies and tools, rta.poker stands as a promising option. For further insights into the world of online casinos, check out 카지노사이트. In this ever-evolving realm of online poker, rta.poker's advanced features and commitment to excellence have positioned it as a noteworthy companion for players seeking an edge in their poker endeavors.

WPN Network on ACR and its corresponding skins like Blackchippoker, Pokerking is now available. We also added functionality to enable 6 tabling, VRAM requirement on the GPU is lowered to 4GB.

Hello, I would like to warn people against using this product. I tried it after finding their youtube video but after playing for some time I was banned by GG. I tried to get my money back from rta.poker but not expecting that to happen. Not getting my GG account or the money there back either. Warning.

This is the first time we hear about someone getting banned and we'd think it would make more sense for a genuine customer to first contact us about this matter privately. Also we'd think it would make more sense to write in the youtube comments (or at least there as well) if you found us trough there but let's not get into speculation/accusations.
So excuse me if we don't believe you to be a genuine customer. Can you provide a discord username/email depending on where you purchased from? If we ever do get a ban we'll be the first to notify users of the increased risk (since after getting a ban we'd consider the risk to be increased, compared to now) however this has not happened yet. Also to be fair we're probably the last RTA getting banned compared to all other products we know of on the market since we do have the best security features in this department (using OCR, custom postflop sizes adjustable by the user so everyone can play with their own sizes, users will soon also be able to choose different preflop open sizes so half of our users might choose 2.5x open or 3x, one will be able to customize even further with for example opening 2.5x from every position except BTN where they open 3x etc etc.).

Please reach out to us with your username/email/license key so we can verify you're a real customer and the reason you were banned. If this happens to be a genuine comment (which we doubt but we'll give you the benefit of the doubt) and we can verify it we'll write a statement here ourselves on the increased risk. Meanwhile more than half a dozen players continue to be using our product each month without any issue on GG alone.