DESTROY Your Player Pool Using This Method
Vložit
- čas přidán 22. 05. 2023
- We live in the era of solvers and GTO strategies. However, poker is a game of people. We must not forget that poker is only profitable because other players make mistakes. It's our job to understand these mistakes and punish them for maximum EV. In this video I show you a study method that allows for finding your opponents leaks easily using H2N.
Get H2N (with discount): hand2note.com/?aid=e30e Use 'pokermadesimple' promo code on checkout for 10% discount on first purchase!
----------------------------------------------------------------------------------------------------------------------------------------
✓ If you are looking for a staking deal, or a coaching for profits deal, send an application at www.metagame.poker;
✓ If you want to hire me as a private coach, send an email to saulo.costa.94@gmail.com;
✓ If you want to get in touch with me about doing other types of business, send an email to business@saulocosta.poker.
----------------------------------------------------------------------------------------------------------------------------------------
- The HUD/Tracker that I use: hand2note.com/?aid=e30e Use 'pokermadesimple' promo code on checkout for 10% discount on first purchase!
- The training platform I use for my studies:
gtowizard.com/p/pokermadesimple
- The software I use for Betsizings, RNG and Tiling: affiliate.jurojinpoker.com/sa...
----------------------------------------------------------------------------------------------------------------------------------------
- Follow me on Instagram: / saulotrcosta
- Follow me on Twitter: / saulocostapoker
- Follow Metagame on Instagram: / metagame.brasil
#poker #pokerstrategy #pokercoaching #onlinepoker #pokermadesimple
Kind of ironic that in this video, all everyone is talking about is the bots I programmed 😆GTO is cool and all, but it shouldn't be an end in itself. There is more to poker than unexploitable strategies! If you would like to go deeper in your exploitative explorations, then H2N is the way to go. Signup using my link to support the channel: hand2note.com/?aid=e30e, then use 'pokermadesimple' promo code on checkout for 10% discount!
I mean, this video is interesting but people can't implement anything without the GTO ranges to compare against. So overall, while educational, it's not practical for anyone who hasn't programmed their own GTO bots to generate hand histories with.
Something I noticed is chess players don't talk about engines like poker players talk about solvers. Chess players are doing it right imo.
@@c0l3slaw As I said in the video, you dont need GTO hand histories for gto data. Anyone with a pio license can run aggregated reports, and if you dont want the trouble you can just buy GTO Wizard and the reports are there
@@saulocostapoker Apparently I didn't understand when your shared that. Either way, I liked the video as it's very educational regardless. I don't see much content like this, and this is the type of thing I've been trying to learn lately.
Kind of obsessed with the idea of fitting poker bots with real hand histories so that you simulate non GTO play
I am a Chinese player, very honored to see the video you made. You are the most sincere and selfless video author I have seen who shared his knowledge, thank you very much. It's a pity that my English is not good, so I can only express my thanks through translation software!
Just such a great video! Easily one of the best and most instructive videos about how to develop your own exploitative strategies IMO. Great work Saulo!
Thanks for the breakdown. Definitely interested in a more detailed video about Step 4.
Very good video Saulo. I noticed when looking at the fold to 3bet statistics, people are sevveeerrreeellyy over defending with calls. Although they under 4bet which should allow us to 3bet wider, when people are defending an extra 10% of hands from BN v BB 3b (which seem to be every pocket pair, suited connector and AT,KQ,KJ), unless they are studied with their real life ranges in PIO, they will have to make some very light defends in some situations but probably will end up overfolding these hands in practice post-flop.
Yeah, and a lot of people just call way too much if not pure with AK instead of
4betting it which also makes the Call3b higher and 4b lower. So be cautious on those boards on mid stakes and below
another great video!!!, I've been studying for the last 3 months with simplepreflop for preflop and working a lot with preflop nodelocks and it's impressive the results that the solver shows from small preflop frequency deviations, strategies can radically change to one side or the other starting from small deviations, very similar to the PIO nodelocks, I will give an example here, BB x BTN with the BTN having an under4bet frequency (very common in the population) the solver even has a 3bet/fold strategy (high frequency 3bet ) and gives up the call for the simple fact of facing less 4bets...
Yeah...that's one of the reasons why I 3bet 30% vs fish! Good work Kako
Love your content man! Great addition to poker youtube :)
Very cool video, Saulo! I think you'd have a market for a programming course built around building bots for the sort of mass data analysis you covered in this video. Will Tipton did a fictitious play bot building course for poker players back in the day and had a small amount of success with it, but I think a course focusing on mass data analysis as the end goal would be more practical and might get a lot more interest. It'd also be >0 sum in all the cool ways. I'd be interested :)
Love the video, def interested in the bot to bot programming part
these videos are next level. well done man
Melhor conteúdo de poker no YT, parabéns demais!
Thanks for the content. Great Video!
You're incredible, thank you so much for share this with us.
Hey Saulo, Thank you for the great content!
Can you make a video/comment about anticipating exploits on previous streets, if is worth it?
For example, If villain has a bit of overcall turn, most combos of bluff become -EV. However, If the same player has ton of overfold River in most runnouts, it becomes +EV to bet/bet Turn and River?
The same with bluffcathing. If villain underbluff a bit on Turn, but a ton on River, call/call becomes +EV always with the previous 0 ev combos bluffcatchers?
If so, it's something to do intuitively, or can be done in Pio, even if it involves deviations on two streets with multiple possible runouts
Great video, Saulo! I really love this idea and just started using H2N. I want to ask you about the definition of reg/fish. Are you using the default definition at "Who is reg" setting?
Love this content . So helpful n yes would be great if u could make a followup about how u programmed bots for the GTO data gathering pls n thanks
So which are the 5 main deviations ( pre and post ) you noticed with this method ? Except under 4 bet, under 3 bet, overfold to probes ? Can you do a video on this topic ?
Hey Saulo awesome video and insight. Out of curiousity is your bot ablle to play other forms of poker e.g plo variations?
Thank you very much for the content. One question, is it possible to evaluate how these 2k regs you put together play against each other only? For example excluding hands where the fish vpips
Saulo, great content! What is the best way to filter for fish’s? Just clicking on that fish box is enough?
Really sick video!
More this content pls. Great video TY
Kct mano, que massa! Congrats, vc é o cara mesmo! To pensando agora em como fazer isso pra MTT, siiiiiiiiiiiiiick! hahahaha
i am a programmer and am curious how you get the bots to play. are you using an api to presolved available solutions like GTOwizard or do you use your own computed solutions for the whole game tree (or is there intensive on the fly pio calculation?)
Keep it up man!
conteúdo bom de mais man! parabéns
Hi Saulo, sorry I have a question that's related to past videos - what is the RNG you use? Is it embedded in GGpoker or something else?
Saulo, how are you? I hope that very good. I have a question. Is there any other way to get HH from GTO game to introduce it to H2N? Because I would like to buy it but I don't know where to get these GTO hands to compare them directly without using 3rd party software like Pio or GTOwizar. Thank you
Hi Saulo, i've a question about MDA, i have made a pop up of 160 stats, now i want to study each of the 26 flops that i choose like a subset. And for some flops stats miss samples, what is the most common choice at this stage : clustering flops or buy more HHs and an harddisk ? What would you do ? Thanks. JF
Hey Saulo, given that you were a part of the GTOWizard coaching team I assume you've been given access to their databases/APIs, which regular people can't access, in order to build your bot? Or actually any user with sufficient coding knowledge could accomplish a similar thing?
Great video, I'd also like to know about the programming. Did this use machine learning, was it using counterfactual regret minimization or a different algorithm/ approach? Really liking your videos so far keep them coming :)
I would also love to see how the bot is made! It would be very helpful to be able to generate large data sets of my own bot hands
Очень познавательно! Браво!
would love to see how you code the bots!!
Could you explain where to find GTO database please?
Like can we actually get them with GTOwizard and put them hand2note just like you did?
Thanks for this awesome video
great video, ty
Is there an easy way to import hand histories from live poker? Or does it have to be code formatted.
Thank you!
yet another great video. what I'm afraid of are AI-powered bots that use player-specific HUD stats and info about pool tendencies (when it doesn't have enough hands from that player) combined with GTO to maximally exploit opponents. that, no human can beat.
I’m sure it’s possible to make, but people would have to want to use it to cheat to kill the games. Which most people aren’t doing with the currently available tools so I’m not too concerned
def interested in the bot to bot programming part
Programming part would be beautiful. It’s probably worth a lot so I’d expect you haven’t open sourced it but I believe you probably have a few more programmers subscribed than you’d expect (cs people love this game)
Yeah I'm surprised by how many people would like to see the programming part! Here is to all the programming nerds out there 🙌
Hey Saulo, really good video!!! So with the data you can say, that population is thighter than they should be (in gto)...so you should basically overfold f.e. vs 3bets of population if you play yourself gto ranges, or not? And if yes, how comes that you made a video that you should not overfold to population, because they overbluff? thx for an answer PS: could you please share the 5bet percentage which the gto bots had? Thx a lot, keep up the great content
Another question: Why do the 3bet calls differ so much from the bots to the population? Is this because population is extreamly overcalling, or is the fold percentage of the bots maybe higher because they used the larger 3bet sizes (f.e. 13 bb in BB vs BU) - ...which isnt used that often by population + there are sometimes to small 3bets - ....and calling gets less appealing for the bots if the sizing is larger? hopefully this is clear how i wrote this haha thx for an answer!!!
Hey, as a programmer myself, i would love to see your bots program :)
love to know more about the bots!!!!
Great video. Would love to know more about the programming part.
Definitely! I'm thinkering with the idea of building a bot that produces HH based on gto strats.. I just don't know how to get those strats to start with.
Yea very interesting, I would like some more info too
Would love to see more of how you programmed the bots
Good stuff
super vidéo !!!
Would love to see Bot making video !
Salve Saulo! Adoraria saber como funciona os BOTS pra gerar as Hand Historys e poder importar pro H2N, jogo MTT e isso com certeza agregaria nos estudos, tmj e parabéns pelo conteúdo de qualidade!
Obrigado Mateus! Isso não está disponível publicamente, eu programei pra fins de estudos pessoais mesmo
@@saulocostapoker Entendi, de qualquer modo parabéns! Bem dahora o approach que você tem, continuarei acompanhando!
Obrigado pelos vídeos, Saulo. Vc poderia informar quais parâmetros inserir em "Base Stats" do H2N para para termos um bom filtro de jogadores recreativos?
Não precisa mexer no base stats. Só selecionar "Is Fish" no topo. O H2N já tem uma configuração padrão para determinar recreativos
@@saulocostapoker obg
12:04 how does H2N differentiate regs from fishes?
But nodelocking based on statistics cand be misleading at some degree. For example tow players may have the same stat value for check/raising ip srp but choosing their combos completely different
Do you sell these generated GTO HH?
Love it
Great video!! Can you explain why GTO folds 60-70% Vs 3Bet and Regs only fold 50%? whats going on?
Are all the regs wrong overcalling 3bets?
Yes, most people are overcalling 3bets. Another thing is that these high folding frequencies of the solver are against the high 3bet sizings of the solver itself (like 13bb 3bet from the BB vs BTN-CO). In real games, most people don't 3bet to sizings as big as solver does, therefore it is correct to call more hands preflop.
Why would the bb/100 and the ev bb/100 of bots differ? Because AI EV does not take care of the rake that will be deducted from winning the pot? If that's true it's funny because it means that everyone is meant to run under EV actually (according to H2N)
When you tried to get the bots out of the research for real people, you used the hands input and set it to " 0 to 75k '', shouldnt it be something like from 9k to ∞ or whatever number so the bots wont be included?
i did something similar to your bots but only for HU pots, did you manage to also make them play multiway postflop?
i also like to let them play my hands (board + runout +my holecards + villains holecards if known - unknown stuff is replaced by randomly picked from range but based on gto frequencies) over and over to see what winrate/stats they would be able to get and compare to my actual results - this also helps to detect rungood/runbad so that i dont tinker with a spot too much if my results are bad but mostly due to flops/runouts or villains hand that were bad for me
I think he also only made it HU only.
@@lukebruce5234 yeah might be, i will try do to the same and just let multiway pots check down to capture preflop and overall postflop HU stats - so far i was just bulk generating single situations like 3BP BB vs BTN
heh turns out when running my own hands through the bot using villains hand when known is not a good idea - makes sense, we only know the holecards that show down and those are very biased towards strong hands - so in that simulation our results are just terrible :)
Could you please show the redline of the bots?
Redline = 0
Blue line = 0
Green Line = 0
............
Bots are playing exact same strategy vs each other.
@@sssupersoak426Actually gtowizard made a video about this like 2 months ago. Gto bots actualyy win in blue line and lose at redline. This is due to multiways, where two player win from a folding player, so in multiway scenarios you expect to win in showdown but lose in nonshowdown.
The name of the video is ”Where does EV come from?”
Yea I saw the video as well , its like triangle shape . Blue line is winning, redline equally loosing as much , so green line ends up 0
My brain just exploded. Haha
I'm still trying to take what I watched and understand it conceptually.
Everyone folds sb and bb a lot, btn wins uncontested, pretty sure red g reen blue the lines are even in that spot
The scenario In the video that I'm trying to conceptualize
Btn raise, sb mostly folds, bb forced to show down (with a weaker range)
The money every bot (sb) loses in this spot is essuntually returned to them on their btn ( via blue line)
I think I'm starting to get it. So the flaw in my logic is using HU to conceptualize a multi way situation.
I would like to see a programming video
Do you use H2N only for MDA? Because if I remember correctly you also used HEM2 in some of your other videos...
I use for everything. HM I use to access old databases
And where can I get the bot's hh?
Did the bots play only 2-way pots? I am aware of Piosolver Python wrapper, did you use that or u've built custom wrapper? Are you H2N hands filtered to exclude Multiway spots?
Hu pots only, as preflop solutions allow no cold calls outside the BB. Ive built my own library to interact with pio
I’d love to see how you built and programmed your bots. Would pay to see how you did
I’ve got a sample on my pool in Holden manager but I’d have no idea how to write a script to simulate GTO wizard in PokerStars hand histories.
What’s the best way to access this sample?
If you can't program a bot yourself, the way to get GTO Data is by running aggregated reports in Pio Solver. GTO Wizard has them as well, but only for flops and turns, so you would have to run rivers yourself
@@saulocostapoker Cheers
Are the stats and popup you use for H2N in this video available for purchase?
Yes, send me an email at business@saulocosta.poker
@@saulocostapoker done!
0.1% 3b HJ vs UTG whats going on with the bots in HJ?
So you're still making good use of your university degree :) I'm also interested in your program :)
Is H2N useful without these reports you created?
I guess we could also buy these reports from you?
Saulo the pool is way underfolding against 3bets so I went and checked all my light 3bets. ALL MY LIGHT 3BETS ARE LOSING! Should we never light 3bet in anonymous pools? This is a bit of a revelation. Would love your feedback here because its confusing me a bit. Thanks
Two reasons why light 3 bet is still ok:
Pool also under 4 bets which makes light 3 bets better. Did you change both overcalling and under 4 betting?
Secondly GTO wizard uses large 3 bet sizes and the pool is facing smaller 3 bet sizes so will naturally call more. So pool vs your correct larger 3 bets they will be folding more than against random smaller 3 bets. However pool does still over call vs all sizes it’s not as bad as it seems.
@@riffin632 I looked deeper into my filter and my offsuit broadways were making my results look bad. Ax’s and sc’s were performing well.
Thanks for your input :)
@@mikegoodwin5951 I had the exact same reaction of surprise from the huge over calling by pool you can see my comment which Saul replied to
How do we get our hands on GTO hand histories?
Gotta generate them yourself...don't think we will see it available publicly anytime soon. Maybe someone will sell them eventually, but likely as a subscription service where you don't actually get to download the hands, just replay them and see statistics
Hey Saulo I applied for Metagame about 2 weeks ago but haven't heard anything? Does that mean I didn't get accepted?
Hey, Bert! Hows it going? Our selective process says we will contact within 15 days. Your application wasnt analyzed yet! We are still on schedule :)
@@zinhao7077 okay thank you!
El segundo punto se consigue con datamining, que puedes tener millones de manos, jugando no se consigue mucho
3:16 yes please :D
WOW You made your own GTO bot? I asked in another video what do you use to drill as it doesn't look like GTO Wizard it seems you're using a program to drill that you made? WOW please show how to do this LOL..
Yeah I made my own custom trainer. I wanted to train in a program where the experience would be similar to grinding, with tables in the desktop, sounds and etc. Couldn't find anything publicly so I coded myself
@@saulocostapoker Genius man, I'm a poker scrub video game player loved your RIO/GTO Wizard videos you're killing it with the channel thanks for the reply..
Can u share GTO BOT hands?
Bold
I would pay for that bot hands jajajaja
@@saulocostapoker At least say they WWSF and WTSD, i beg u
I dont know why GTO fold like 66% pre vs 3b. I mean, i can literally 3b anytwo with those stats no?
No...first of all most of the time when you 3bet there will be people left to act that can punish wide 3bets with cold 4bets (except BB of course). And second, its all a function of the betsizes used. 66% fold doesn't mean anything before considering sizes used. If HJ opens 2x and you 3bet to 12.5 from BB, which is close to optimal size, you are risking 12.5 to win 2.5. You can only profit from any 2 cards 3betting if HJ folds over 83.33% of the time
Am I tripping or do those bots have like a 67% fold to 3bet. Despite the bots 3betting wider than humans.
Does this mean the correct exploit would be like a 70% fold to 3bet. Surely not. I only have like 50% fold to 3bet I thought that was right.
Fold to 3bet% is a function of the sizes your opponents use when 3betting. My bots played with GTO Wizard Simple ranges and sizings, and there the SB and BB 3bet to big sizings, against which its optimal to fold 65%+ most of the time. If people in your pools are 3betting to lower sizings, naturally you will want to call more. But from my experience, everyone is overcalling vs 3bets in most pools, so you are likely doing that too
@@saulocostapoker Yeah I figured it would be those two reasons but still wouldn’t have expected such a gulf big eye opener to a leak I just realised I had.
I’m gonna start trying to be a bit more disciplined especially as population under like I’ve just seen KQs LJ vs SB is a mixed fold which is absolutely something I’ve never done in 500k hands of poker despite the correct adjustment to probably be to fold more.
@@riffin632 You could also run a filter for the mixed strategy hands and see if you are protiable when you call a 3bet with them. You would just need to be losing less than you open raise size. So if you open 2.5bb and call a 3b with KQs you would need to be losing less than 2.5bb for it be be a profitable call.
Hey Saulo! At 14:29 you say humans don´t 3b enough. I wonder, wouldn't that be the right way to exploit a pool that under folds vs 3b? I see your bot folds much more against 3b then the population.
It could in fact be a good adjustment of the population agaisnt itself. It doesn't make it less exploitable though...someone who knows this can just open more often and fold more to 3bets
@@saulocostapoker thanks! You're the man!
Can’t believe this was free
Next week video will blow your mind then...
would you share bots hh?
Unfortunately, no, I won't share that
@@saulocostapoker understandable, nice videos as always, keep it up!!
Help me please
great vid! would you be able to make those bots play exploitively against the human profile you have?? would be interesting to see how much winrate/WWSF/redline it would get
Were you ever a part of poker detox?
Not really, although I recorded some strategy videos for them in 2018/2019. That being said, I was heavily influenced by Nick Howard in the beginning of my career, as he was the only one talking about generating exploits through analysis of the player pool, back in 2016-2017
@@saulocostapoker cool. Yes the theme of this video reminded me of his mass data approach- looking for exploits in the fields
absurdo!
Poker is a zero sum game. Poker coaching is a scam.
Poker is a negative sum game because of the fees to play.
However, there are plenty of professional players with enough skills to make a good living!
would you contact me regarding gto auto play. appreciate it