Finite-Shoe Research Workbench
Experimental analytical utility. Not a live-play adviser, casino assistant, betting recommendation or profit predictor.
Verification warning: Exact recursive split enumeration can exceed its state budget on full-shoe configurations. When that happens, use the Monte Carlo mode only as a fixed-policy approximation. Its confidence interval measures sampling error, not policy/model error. Results are not certified for casino play.
Configuration
Results
Choose a configuration and run a calculation.
Methodology and limits
- Rank composition is exact: Ace through 9, plus one ten-value bucket.
- Exact mode tracks card removal across all split hands and resolves one shared dealer hand after player actions.
- Exact mode throws when the state budget is exceeded; it does not silently fall back to replacement sampling.
- Monte Carlo mode shuffles a finite shoe for each round and uses a fixed basic-strategy heuristic, not statewise EV optimization.
- Insurance and surrender are not included in this split-only workbench.
- Physical-device, browser-matrix and accessibility certification remain outstanding.