Jakob Tøstesen

Thanks for another great video!
I made a comment in your previous video about the potiential use of RTA (https://www.runitonce.com/poker-training/videos/nuno-alvarez-stefan-vs-linus-exploitative-play/) which I am quite curious to hear your opinion on :)

Hi Nuno

Great video series. Very interesting discussion of the hands. It seemed to me that Linus is playing quite a bit more solid than Stefan during this particular match.

With regards to your discussion on RTA in the end of the video. Given how fast the new AI feature of GTO Wizard can create solutions for any customized spot (within a couple of seconds as mentioned in this video https://www.youtube.com/watch?v=mi5F1-69QOw&t=624s), I find it hard to overlook the longterm challenges which this presents to online poker.

In a article by GTOWizard they write:
'How accurate are these solutions, you ask? Very. We’ve accomplished speed and accuracy that outperforms all benchmarks. We decimated the ACPC champion Slumbot for 19bb/100 in a 150k hand HUNL match, and averaged a Nash Distance of only 0.21% pot when nodelocking these solutions against PioSolver.'

If a HUNL player has access to a software similar to AI GTOWizard, and has some external screen capture software which automatically inputs all the spots in real time, and the player just play according to the solutions presented by this GTO software, how would it be possible to beat this person in a HU NLHE match?

From seeing the hands in the video series it seems to me that even a player like Stefan is making a fair share of blunders. Given the stakes that is still being played online there must be a huge money incentive for technical people with enough money to develop a tool like AI GTOWizard with fewer anti-cheating limitations.

A counter-argument I have heard from Phil Galfond and Saulo Costa with regards to this threat, is that pokersites of course are incentivized to avoid cheating going on, by also investing big sums of money in AI detection which can help catch players who try to cheat with RTA.

But lets say someone were to develop something which I suggested. In order not to get catched I assume an easy solutions would just be to occasionally deviate from the RTA's / GTO's suggestion, in spots where the EV cost would be minimal.

Very curious to hear more of your thoughts about this. Let's say Stefan were to play against GTOWizard's AI software. I find it hard to imagine that AI GTOWizard wouldn't crush Stefan and many other top regs quite hard in the long run?

This was not course not possible previously with PIO where running a simulation takes quite a long time even with a very powerful PC/server. But since GTOWizard has now proven that it is already possible to solve any custom spot in just a matter of seconds, how is this not significant threat to online poker, and especially online highstakes HUNL?

Great video as always. Appreciate you for sharing.

Just watched the last hand. Excellent breakdown, very interesting.

Loved the video!

Great material, thank you.

In addition to what Francesco mentioned, IP will also fold overcard combinations, which is pretty significant.

Hi Francesco,

First of all thank you for the great video.

In your first simulation (1/3 flop, 2/3 turn), in addition to the concepts you mention, I think an important reason for KQ pure betting and AK mixing might be, that KQ blocks a very significant portion of IPs bluffing range (QJs-Q9s).

So.. when checking AKo IP will be bluffing significantly more compared to when checking KQ.

What do you think about this?

Luke Johnson, you are freaking awesome!

Awesome format, loved it very much!

Thank you for the great content. A lot of inspiration, as always.

In you work with the sim, I think you overlooked the fact that KK-99 all mixes calls against you river jam. Especially the combinations of queens that block QJs mounts to a significant number of combos.

Hi Teuness,

Thanks for the video, cool format.

I have just watched the second hand and I was a bit surprised about the flop cbet sizes you have chosen. From my experience, Pio likes the 50-60% pot sizing on the flop on most J-T high boards, so I was wondering why you went with 30% and close to pot. What is your reasoning behind this choice of strategy?

I ran the hand in Pio with near pot and 50%, and Pio uses the big sizing 31% and 1/2 pot 46%, while checking the remaining 23%. Interestingly, this changes the the strategy on future streets dramatically, which is very important. In your exact hand, Q9s have a preference for the 1/2 pot size on the flop, and given the runout Q9s blocks both turn and river with a very high frequency.

So, what Im trying to say is, that we need to be aware that the choices on the flop very much defines our strategy in the rest of the hand.

Thank you for the excellent content!

Great content. Thank you.

Thats for the video, great stuff.

Nice video :) What is your bb/100 on NL500?

15.30 in with 66 - Do you really think that A4 and 55 are going into his overbet range on the turn? According to my sim A4 and 55 are never overbetting the turn.

5 min in with the Jd9d hand where villian raised river. Did you check what he had?

27.20 on the 3h turn what is the logic behind btn checking a lot here with various sd?

Great format and analysis!

Nice video. Anyone who have investigated what PIO likes to bluff with in the last hand where villain had KK?

Did you PIO what the pot likes to bluff? :)

Really nice video Mark!
Think it is something that would increase your overall winrate a lot if you implement it correctly, also since it is not used that often in practice, so might be tougher for opponents to play against. Also noticeable that we see a pretty clear pattern of which hand categories to check raise on most boards(which you did a really good job pointing out btw). The hard part ofc is to get the frequencies right, but definately something i would try to implement.

Thank you for the good answer.

Hi Francesco,

Great vid. I think you cover a lot of important topics in your analysis.

Still, I think you fail to mention a crucial factor in your comparison between the range C-bet strat and the polarized C-bet strat - the loss of EV stemming from not being able to replicate the more complicated strat. Furthermore, this issue even accelerates when getting to the turn, when the three grows exponentially leading to more mistakes being made by OOP in practice.

What do you think about this?

Hi Saulo,

First of all, congrats on you son :-)

You mention that 1/3 potting with a high freq on the flop isnt printing versus good regs at mid-stakes+ as it previously did. I agree with that. But what do you think about the node 1/3 potting flop at a high freq and barelling the turn. This node is less studied by OOP than the flop for obv reasons. Do you see any exploits here? Will OOP overfold the turn instead of the flop?

At 7.40 you argue that you would flat AKo versus a 3bet MP versus BTN if the pot had been HU. Can you elaborate on this? Pretty sure the solver suggest to 4bet here with a very high frequency.

As always very high quality content! Thank you!

Great vid! Especially loved the Pio analysis.

Great stuff. More of this type of teoretical content would be much appreciated :-)