Posted 10 years ago
My recent video on GTO got posted to reddit and there were some interesting responses, I thought it would be good to have a thread to discuss the points raised as there are a lot of differing opinions on exactly how useful game theory is in a real game of poker. I'll cross post some things from the reddit thread here and if there's anyone here that has any questions or points they want to raise that would be great.
Posted 10 years ago
[url]http://www.pokervip.com/school/poker-coaching/videos/view/223/[/url]

The video!
Posted 10 years ago
1 reply from reddit:
donkeyotee said:
"An optimal set of range, bet sizing, lines, and frequencies will be almost impossible to model. There are too many "it depends" solutions for optimal play. This is because we play against people. Maybe in a heads up situation it might be possible to get close enough that a system would win over the long run against any individual on the planet but as more people become involved and their actions become more and more dependent on not only the actions of other players but on their perception of those players, how they think they might react to certain situations and how they think those players think they might react the game starts the get to a level that I don't think any system could hope to defeat the game alone and without human decision making.
Anyway, this brings me to my game theory for beating poker. Identify the worse players as quick as you can and play them as much as you can for as much as you can afford to lose. Don't worry about trying to level the better players at the table just try to get it all in per flop against regs with pocket aces or what ever they are willing to put money in with worse than you have so you dont have to play the. For as much post flop. Just isolate a catch fish. That's it."
Posted 10 years ago
And my reply:
" In chipEV situations, 'people' aren't the reason GTO "won't work", we just don't have a GTO solution to poker due to the sheer complexity of the game.

Your second paragraph is of course good strategy, however I disagree with the "That's it" part. I read once that 80% of poker hands are played by 2% of the players[1]. This means that the vast majority of your opponents are going to be regs. Also, because poker is so complex, everyone at the minute is still playing terribly, in absolute terms. If some alien intelligence landed tomorrow and planted a true GTObot in NLHE games, it would crush everybody. Poker is constantly evolving via a process similar to natural selection, i.e. some people go broke, some people make money and stick around. It will constantly evolve, and strategies will rise and fall along the way.

Example: 5 years ago you could easily make a comfortable living with the strategy of playing tight most of the time, but isolating fish and making many small bluffs at them. Then, finally, you show up with the nuts the time when they were ready to call all in. You're right that a GTObot would not do as well against this fish, it would miss out on some of the small pots and not build up as bluffy an image to set up the big payoff, and then it would be bluffing some percentage of the time when the human player would know that the fish is "ready to blow" and is never folding. Obviously don't bluff when you someone is not going to fold.

However, nowadays people that play like that get exploited by other regs, and struggle to find enough fish to get by. They're having to bumhunt harder, maybe waiting at heads up tables and not playing anyone that sits with 100bbs, or playing lower stakes on smaller sites. Knowledge of GTO would help them adjust their strategy to stay ahead of the games. Doing that, they could still be making good money off other regs. E.g. what do you think would happen if a winning 2knl reg dropped down to 100nl for some reason, maybe a prop bet, and played full time in those games without any bumhunting or game selection of any kind. Personally, I think they would win comfortably, because the 100nl regs will be making many mistakes that the 2knl reg is used to exploiting, and due to his greater experience the 2knl reg will be able to anticipate their 1st adjustment, both the adjustment and the timing, quite accurately as well. As I said, the 2knl games are also exploitable in the same way, just nobody has got there yet. To use an example from outside of poker, imagine if Andy Murray could go back in time and play Fred Perry at tennis, I think it would be very one-sided, but back in Fred Perry's day people would have said "you'll never make any money playing Fred Perry at tennis, don't bother", when in reality it was possible to be much better than Fred Perry with a lot of work.
I'm rambling now, but I think the point was that I think it's a very relevant and fruitful area for learning.

[1]Source: Nate Silver's 'The Signal and the Noise'"
Posted 8 years ago
I haven't watched the video yet but have been thinking about GTO poker recently but been wondering wetehr this does exist. There are people significantly more intelligent than me who say that it does so I assume the are right.

However my thughts so far are that against TAGs and looser players we can have a strong enough range that we can play a particular strategy that is balanced and strong enough. But if we are up against a total NIT then if we don't adapt our strategy then we make it to the river where the biggest bets are going in with a significantly weker range than our villian. Therefore we would need to significantly tighten our opening range loads to compensate.

If anyone can enlighten me that would be great.