Online Poker Strategy: Wrapping Your Head Around Game Theory

Jason Lake

Tuesday, February 24, 2015 2:15 PM GMT

Being able to understand and apply game theory is what separates today’s big winners at the poker table. But no matter what level you’re at, you can add at least some game theory to your poker strategy.

It’s all over for the rest of us: The computers have solved poker. Well, heads-up limit hold ‘em poker, at least. Researchers at the University of Alberta reported last month that they have “weakly solved” this particular game using a computer-generated algorithm called CFR. It doesn’t mean the computer would win every hand, but over time, it would win more using this algorithm than with any other possible poker strategy.

Good thing most of us are playing no-limit hold ‘em. It’s going to be a while before the number-crunchers figure out how to deal with all the possible bet sizes involved in NLHE. But if you really want to clean up at the online poker tables, especially with the limited number of “reads” available to you, learning how to play unexploitable poker like our computer friend is the way forward. That means game theory. Hold still, this will only hurt a little…


Undefeated in Sixth Grade
Let’s start by looking at a much simpler game: tic-tac-toe. Did you manage to solve this game when you were a kid? Even if you’re the one going second and putting the ohs on the board, you can always make sure that you never lose, no matter where your opponent puts the exes. Then, with any luck, your opponent will screw up and give you the opportunity to win.

That’s the basic idea in poker, too. By using ratios and probability and other math stuff, you can build an unexploitable poker strategy that, at worst, should break even in the long run, no matter how good your opponents are. Then, when you happen to run into opponents who make mistakes – and we all make mistakes – you can do even better than break even. If you can grasp that idea, you’re already on the right path. No advanced degree in game theory necessary.


I, Poker Bot
You also don’t have to do all the heavy math to figure out these ratios, either. Other people already have, and they’ve made millions of dollars at the poker table over the past 10 years or so. You can even buy software these days that does pretty much what the computers at the University of Alberta have done, although there’s quite a way to go before they come up with the “perfect” algorithm for NLHE.

However, it’s always better to know why you’re implementing certain poker strategies, rather than just following them robotically. Building an unexploitable strategy starts by looking at the very basic actions at the start of a game. Let’s say we’re playing 6-max NL100. The blinds are 50 cents and a dollar. If the button opens, the players in the small blind and the big blind need to defend with a certain frequency (either by calling or raising) in order to prevent the button from just stealing every time and profiting.

Play some Poker Today with 2015 SBR Bash Poker Tourney

Tree Fiddy
That frequency depends on how big the button’s opening bet is. If it’s a larger bet, say $3.50, the blinds should defend less often than they would if the button min-raised to $2. When we talk about frequency here, we’re really talking about which hands you should defend with. If the math says you should defend your big blind 50 percent of the time in this spot, that means (all other things being equal) that you should defend right now if you have a hand that’s in the Top 50 percent in terms of strength, not that you should flip a coin.

Once this fundamental piece of poker strategy has been absorbed, you can build from there and figure out the “Game-Theory Optimal” play for everything that happens at the table. Correct that: You can figure out something close to a GTO strategy. Again, we’ve got a long way to go before we get there, and even the top players in the world disagree somewhat on how to apply GTO concepts to poker. So there’s still hope for us mere mortals.