How AI Changed The Game of Poker
We want to share a recent discovery that we made last year, and it opened our eyes to a level we never thought possible in online poker games. As we dive into the world of AI and solvers in poker, join us. This post will look at what we thought was a simple spot in No-Limit Texas Hold’Em. We can safely say that this post completely transformed our view of the game.
A Bit of Poker History
Before we get to the meat of the matter, let’s briefly discuss the history of online poker game strategy. When we first started playing, the best way for us to learn the game was to:
- Read Poker Books
- Mathematical and Game Theory
- Compare strategies with well-known players
These are good ways to learn if you’re beginning the game. It was all based upon theory and history. There was always uncertainty when you were creating a strategy, and it was impossible to determine if your strategy was the best.
You could use simple math and probabilities to calculate expected value for any given action. But you can’t predict whether your strategy will work in real life. You could only try it once and then analyze the results to see if you can adjust.
Online Poker Game Meets AI and Solvers
We now have powerful tools to run simulations and test our strategies in seconds or minutes, depending on how much computing power we have. Solvers can analyze all actions at a spot and determine the best strategy, and they can run simulations that test all scenarios within the constraints you set (decision tree). This allows us to discover mistakes in our strategy that would otherwise be difficult or impossible to spot without a solver.
The Strategy
This brings me to the major flaw in my No Limit Hold’Em tournaments strategy. Our strategy for playing out the Small Blind after everyone folds becomes a Small Blind against Big Blind situation.
It was traditional that you should raise or fold your hands and avoid calling the SB. Flat calling is possible in certain situations, but it’s generally a fold or raise.
The SB solver never folds. It either RAISES (flat call) or LIMPS(raises). Sounds crazy? It’s not.
Let’s compare my old strategy to the solver’s one.
(Red = RAISE; Pink = FOLD, or RAISE; Green = CALL; Blue = FOLD)
The Obvious
The obvious difference is that the solver warns us not to open the fold from the SB if it’s being folded in tournaments (or cash games with antes). The solver will play every combination as an open raise or a limp. When you see the solver’s range asks you to play, it becomes obvious that we were being too strict. This strategy was much easier to remember, and it is not the best.
Balance is everything
It’s not obvious, but it’s the fascinating part of the solution. The solver uses a polarized range to raise the linear ranges that humans have used for many years. Although it isn’t as easy to remember or intuitive, it makes it virtually impossible for an opponent to exploit you when using a perfectly balanced range.
The solver selects strong combinations and makes a call, and the solver also selects weaker but still very viable combos to raise. This means that you don’t give any information when you call or raise, and you’ll have strong and weak combos as part of your range in either case.
Conclusion
Modern AI and solvers have made it clear that many of the things we used to do for years are no longer correct. The ability to solve complex problems makes modern computer solvers so much more powerful than anyone who can use math and game theory.
Modern computers can run through all possible scenarios reasonably and provide a definitive answer for each spot. Sometimes, this creates situations where the solver strategy is too complex for a human to master. The most important skill you can have is simplifying these strategies and making them more understandable for humans.
This was just one example of a straightforward preflop spot in Hold’Em. You can see how complicated it can get for more complex scenarios with more players. Today, the best players use solvers to learn, and those who can best apply these strategies will be the ones who stay ahead of the game in the future.