Ask any curious player the question "is poker a solved game" and you'll get answers that range from emphatic "yes" to a thoughtful "not really — and here's why." As someone who has spent years studying game theory, building strategies at the felt, and testing computer solvers, I want to walk you through what “solved” actually means, the breakthroughs that changed the landscape, and what it all means for real players trying to improve their edge.
What does “solved” mean in poker?
The term "solved" has precise meanings in game theory. At a high level there are three useful categories:
- Strongly solved: For every possible legal position, the perfect play is known and the outcome can be predicted with certainty.
- Weakly solved: From the initial position, perfect play is known — the game is solved starting from the beginning but not necessarily at every possible branch.
- Essentially solved: No exact perfect-play solution is possible, but algorithms produce strategies indistinguishable from optimal play for practical purposes.
These definitions matter. Chess, for instance, is not solved at any of these levels; checkers was strongly solved to a draw by exhaustive search and retrograde analysis. Poker, unlike chess or checkers, is an imperfect-information game — players have private cards — which changes the computational challenge entirely. That privacy creates hidden states and strategic deception, making the problem far harder.
Key breakthroughs: what computers have actually solved
There are notable successes in the world of computational poker:
- Kuhn and other toy games: Very small imperfect-information games have been completely solved for decades and serve as important theoretical stepping stones.
- Heads-up limit Texas Hold’em: In the early 2010s, researchers produced essentially solved strategies for this variant — a major milestone. The algorithm known as "Cepheus" demonstrated an effectively unbeatable strategy for heads-up limit Hold’em by reaching almost-zero exploitability.
- Heads-up no-limit and multi-player no-limit: Advanced systems like Libratus (2017) and Pluribus (2019) used abstraction, self-play, and real-time re-solving to beat elite human professionals in heads-up and six-player no-limit Hold’em respectively. These systems are breakthroughs in AI but are not “solving” the entire game the way we use that word in game theory.
Each of these advances moved the frontier forward. But there’s a critical distinction: producing an invulnerable, exact strategy for ALL possible positions in full-scale no-limit Texas Hold’em is beyond current resources. Instead, modern programs compute near-optimal strategies using smart approximations.
Why full-scale poker remains unsolved
There are three intertwined reasons poker resists a full solution:
- Immense state space: No-limit games allow a huge continuum of bet sizes, and cards and sequences create astronomical numbers of game states. Exhaustively mapping every position is computationally infeasible.
- Imperfect information: Hidden cards mean strategies must account for beliefs and deception, which increases complexity relative to perfect-information games.
- Human adaptability and exploitation: A strategy that is “optimal” under one model can be exploited by adversaries who deviate. True game solving needs to be robust to human creativity and changing meta-strategies.
Put simply: even the most powerful solvers rely on abstraction (grouping similar situations), sampling, and continual re-solving. These are practical and enormously effective tools, but they are not mathematical proofs that the entire game is solved from every possible node.
What the AI milestones actually mean for players
When Libratus and Pluribus defeated top professionals, the headlines read like a definitive end of human dominance. In reality, those systems validated that near-optimal, unexploitable strategies can be engineered for complex imperfect-information environments. For players, the practical implications are rich:
- Study GTO concepts: Game-theory-optimal (GTO) thinking — ranges, balance, indifference — gives players a sturdy foundation. Solvers taught us which spots require balance and which spots reward exploitation.
- Use solvers thoughtfully: Tools like equilibrium solvers are best used as study aids. They reveal core lines and tendencies but should not be treated as one-size-fits-all prescriptions in live games with evolving opponents.
- Exploitative play is still crucial: Against imperfect humans, deviating from GTO to exploit specific weaknesses often yields higher win rates. The practical goal is adaptive mastery, not slavish imitation of a computer.
Personal snapshot: learning from solvers at the table
I remember a late-night cash game early in my career where I stubbornly defended a mid-position range against a loose-aggressive opponent. A solver I'd recently trained with suggested folding more hands in that spot. Instead of rote obedience, I tested the change: folding marginal hands reduced variance and gradually increased my win rate because it neutralized his frequent bluffs. That combination — studying solver output, testing it live, and adapting — is the pathway most players should follow.
Practical tools and techniques for improving
Whether you play micro-stakes online or sit at live tournaments, these practical steps help translate theory into results:
- Learn range construction: practice building preflop and postflop ranges by hand before consulting solvers.
- Study key spots: focus on common high-leverage situations like 3-bet pots, turn play in single-bet pots, and multiway pots where exploitative lines shine.
- Use solvers as teachers: run simulations, then force yourself to explain why certain lines make sense in plain language.
- Record and review sessions: nothing beats feedback. Compare your lines with solver recommendations and track long-term outcomes.
The human factor and the living game
Poker is a social game. Psychology, bankroll management, table dynamics, and timing are as decisive as technical correctness. Solvers can provide a backbone of rational strategy, but they cannot replicate the nuance of tournament pressure or the emotional swings of a deep-stacked confrontation. That's why top humans continue to win: they combine rigorous study with situational intuition.
Where to find further resources
For players wondering "is poker a solved game" and looking for practical next steps, online communities, training sites, and research papers are invaluable. For a quick starting point — and to see how contemporary sites present strategic content — consider exploring is poker a solved game. If you prefer academic reading, search for works by Michael Bowling, Noam Brown, and Tuomas Sandholm; their teams produced many of the breakthroughs mentioned above.
If you’d like to keep learning, try a mixed practice routine: study 30–60 minutes with a solver or video, then play focused sessions where you deliberately implement one new concept. Over months, this iterative loop yields the highest returns.
Final verdict: nuanced, not binary
Is poker a solved game? The short, honest answer is: not in the absolute sense. Some simplified variants and specific heads-up limit formats are essentially solved; state-of-the-art AIs can dominate in both heads-up and small multiplayer no-limit contexts through sophisticated approximations. But full-scale no-limit poker with real-world variability remains beyond a formal solution.
That should be good news. The game remains rich with discovery. For students of poker, the frontier is not whether the game is finished, but how you adapt solver insights, psychological skill, and live testing to gain a durable edge. With the right blend of study and experience, you too can climb the ladder — even while the theoretical landscape continues to shift beneath your feet.
About the author
I’m a player and researcher with years of experience using solvers at the study table and applying those lessons in live and online play. My approach emphasizes understanding principles deeply, testing them under live conditions, and refining strategy based on results and evolving AI research. If you want a guided study plan tailored to your game style, I can outline one based on your typical stakes and formats.
For a concise resource that addresses public curiosity around "is poker a solved game", visit is poker a solved game for an accessible starting point.
