In chapter 20 of Ace on the River, Barry Greenstein explains how game theory helps poker players think about incentives and adjustments, but argues that practical poker success comes more from judgment, context, and psychology than from formal calculations.
What Game Theory Is and What It Isn’t
Greenstein defines game theory as a way to analyze decisions where outcomes depend on what multiple people choose. He pushes back on the idea that being able to compute “optimal” strategies is what makes someone an expert. Instead, he says game-theory thinking is most useful as a conceptual framework for understanding why certain approaches work.
Two Kinds of Game Theory in Poker
He separates poker-relevant game theory into two categories:
-
Cooperative game theory: how players’ choices affect the long-term health of a game, especially keeping profitable lineups running.
-
Non-cooperative game theory: how to maximize profit within a single hand against opponents who are trying to do the same.
Cooperative Strategies That Keep Good Games Alive
Greenstein lists behaviors that help preserve a lucrative ecosystem:
-
Avoiding hit-and-run behavior that makes others quit or copy you later
-
Not antagonizing losing players, even if it might tilt them short-term
-
Not turning seat selection into a public battle that makes recreational players uncomfortable
-
Being willing to compromise on stakes and game choice so others keep showing up
-
Not trying to “break” a key action player in one sitting if it risks driving them away
-
Not acting possessive or insulting by trying to control when and with whom weaker players participate
His point: long-run profit often comes from protecting the game, not squeezing every dollar immediately.
Non-Cooperative Play and the Logic of Bluffing
At the center of non-cooperative game theory is bluffing frequency: you must bluff sometimes so opponents can’t comfortably fold to your value hands. But Greenstein says real poker can’t be reduced to a clean “randomize here” formula because every hand has context—table dynamics, history, and live information.
What Actually Determines When to Bluff
He explains that good bluffing decisions depend on factors like:
-
the specific opponent and how fold-prone they are
-
what ranges and hand strength you credibly represent
-
whether the opponent is emotionally or strategically in “calling mode”
-
your chances of winning without bluffing
-
how much you gain if the bluff works
-
the risk of getting counter-bluffed
-
how your own cards block or enable your opponent’s likely strong hands
-
your table image and how often you’ve been caught
-
situational pressures (money, embarrassment, tournament timing, pay jumps, short stacks)
He notes the same kind of opponent- and situation-specific thinking applies when deciding whether to call a bet that might be a bluff.
Pot Odds Are Not the Whole Story
Greenstein acknowledges that players may use pot odds as a starting point for bluff-catching, but warns that earlier betting often makes it far more likely you’re beaten than a simple “needs to bluff X%” calculation suggests. With experience, players develop a feel for realistic bluff frequencies rather than forcing math onto messy real situations.
Advanced Poker Is “Situation,” Not “Hand”
A beginner asks what to do with a particular hand. An intermediate player adds the betting. An advanced player thinks in terms of the entire situation: opponents, history, perception, and the strategy they’re building across many hands. In that sense, hands are not isolated events—they are connected moves in an ongoing contest.
Computers vs. Humans
Greenstein suggests a computer could be built to play strong, theory-driven poker because it wouldn’t tilt or show tells. However, he argues that exploiting weaker players—where much real profit comes from—depends on “poker sense” that is hard to formalize and program well.
Core Message of the Chapter
Game theory is valuable for understanding balance and incentives, but poker is too complex for a single solved strategy. Greenstein’s takeaway is to use game theory as a guide, then make real-world adjustments based on people, patterns, and context—while also behaving in ways that keep profitable games running.