Methodology
How JustShove recommendations work
JustShove is a fast educational tournament poker tool. It does not run a real GTO solver. The MVP uses deterministic heuristic logic to turn common tournament inputs into a clear shove, fold, call, or raise recommendation.
Not a solver
No simulation claims
The current engine does not compute equilibrium strategy, exact range equity, or Nash charts.
Deterministic
Same spot, same output
The same inputs produce the same recommendation, which makes the tool predictable for study and review.
Tournament-first
Context matters
Stack depth, position, fold equity, antes, and ICM pressure are treated as part of the decision.
Inputs the heuristic considers
The engine scores the spot using broad poker factors rather than exact solver outputs. Each input nudges the recommendation toward aggression, caution, calling, or folding.
- Hero hand and broad hand category
- Position at the table
- Effective stack size in big blinds
- Number of players
- Tournament stage
- Ante state
- Previous action before hero
How confidence works
Confidence is not a probability that the action will win. It is a measure of how clearly the heuristic score clears or misses the internal threshold for the selected action. Close spots receive lower confidence. Obvious short-stack shoves, premium hands, and clear folds receive higher confidence.
How ICM pressure is estimated
The MVP uses simple tournament-stage adjustments. Bubble pressure and final-table spots penalize marginal gambles because bustout risk and pay jumps matter more. Early, middle, and heads-up stages apply different pressure assumptions.
When not to trust it blindly
Use JustShove as a study aid, not as guaranteed advice. Real decisions can depend on opponent tendencies, payout structure, blind format, stack distribution, rake, future game considerations, and exact ranges. The tool is designed to make tournament concepts easier to review, not to replace judgment.
What can improve later
The architecture keeps the heuristic layer isolated so stronger strategy sources can be added over time.
- Precomputed Nash and push-fold charts
- Solver API integrations for selected spots
- Range-vs-range equity estimates
- AI-assisted explanations that cite the underlying recommendation inputs