Pokercat Guide
Solver Overview
The strategy panel is the heart of the Dashboard: once the table is captured, Pokercat rebuilds the hand state in real time and asks a strategy engine what to do. This page explains how the engines fit together and how to pick a mode. For details on each engine, see GTO & GTO+, Exploit and LLM.
How a decision is made
Every time the action is on you, Pokercat runs the same pipeline:
- Read the table — OCR turns the video feed into a structured game state: cards, stacks, bets, positions.
- Rebuild the hand — the game-flow engine reconstructs the full action sequence (who did what, on which street, for how much).
- Query an engine — the rebuilt state is sent to the strategy engine selected by your mode.
- Show the answer — recommended actions with frequencies and sizings appear in the strategy panel, usually well under a second.
The same pipeline covers all streets from preflop to river.
Strategy modes
The mode dropdown at the top of the strategy panel decides which engine answers. Pick one mode and Pokercat handles the rest.
| Mode | Engine behind it | When to use it |
|---|---|---|
| Auto | GTO library, with LLM as fallback | The default. Uses the GTO library whenever the spot is covered and hands the decision to your selected LLMs when it is not (for example some multiway pots or unusual lines). |
| GTO | GTO library only | You want the pure equilibrium baseline, nothing else. Spots outside the library show no answer instead of falling back. |
| GTO+ | GTO library + exploit shift | The GTO baseline with a profit-oriented shift applied on top, driven by the opponent profile you assigned. |
| LLM | Large language models | Every decision is answered by the LLMs you selected, including spots solvers traditionally struggle with. |
| Exploit | Opponent-profile modeling | Maximum deviation from equilibrium against a specific player type. Currently in development — see Exploit. |
Reading the output
| Element | Brief |
|---|---|
| Action bar | The recommended actions (fold / call / raise sizes) with their frequencies. Mixed strategies show more than one action. |
| Range detail | The full range view behind the recommendation, so you can see how your exact hand class fits in. |
| Engine tag | Which engine produced the answer — useful in Auto mode, where the path can switch between GTO and LLM within a session. |
Every decision, together with the engine that produced it and the eventual result of the hand, is saved to your Data Hub automatically.