Quote:
Originally Posted by Menace ll Society
sorry, jorge, but your last sentence is just flat out incorrect
Quote:
Originally Posted by Menace ll Society
sorry, jorge, but your last sentence is just flat out incorrect
Jorge, NL hasn't been "solved"<> bots are necessarily unprofitable in NL.
If you've got a good case to make for your assertion, I'm interested.
As for the idea that bots can't determine when it is profitable to call a hand that may be a bluff:
Even if that is true, that is no guarantee that the bot would be unprofitable.
But I doubt that a sophisticated bot would be unable to handle bluffs.
Example:
A bot's initial programming could have it 3x raise half of all C-Bets made from late position. After playing thousands of hands, the bot may determine that it is profitable to do this only when (these values are arbitrary) the C-Betting player has a PRF above 15 and a c-bet percentage above 60.
As more hands are played, additional refinements and expansions are possible: What is the impact of the bot's hand strength on profitability? Of the opponent's stack size? Does it matter if the opponent's C-bet number is higher during this session than it has been during previous sessions? Do the stakes matter? What if the bot's raise is 2X? What if the bot floats instead?
Many trials would be needed to reach this level of granularity, but once a statistically significant sample size is reached the bot will have a sophisticated decision tree for handling c-bet bluffs, and it will follow that decision tree perfectly.
In the meantime, by simply re-raising likely C-bet bluffs frequently, the bot would be ahead of most micro-stakes players.
A very similar process could be followed to program the bot to analyze missed-flush river bluffs, checked-down on the flop and turn bluffs, and other common bluffing opportunities.
I should note that although I'm in IT, I'm have no expertise in AI. And I'm a fish. So if I'm incorrect, I look forward to your counterarguments.