How Scoring and Probability Work
The game uses a uniform sampling of complete hazard subsets satisfying the square-difference constraint. When you reveal a safe tile, the current gross value is divided by the exact posterior safe probability.
- Initial quoted value: 97% of captured stake.
- Probability 1 reveals: Leave the gross value unchanged.
- Other reveals: Multiply the current gross by the inverse of the posterior safe probability.
For example, with 3 hazards, the support count is 92. If the printed label is 3, the posterior safe probability is approximately 0.8369565217391305. With a stake of 20, the initial gross is 19.4. After one safe reveal, the gross becomes approximately 23.179220779220778.
Wallet, Balance and Restoration
The demo starts with a 1000 virtual coin balance. Separately, manual restoration is available while idle: if your balance drops below 10 while idle, you can restore it to exactly 1000. This is not an automatic refill or a deposit.
- Fractional carry: Persists in browser localStorage across settlement, refill, and reload.
- Unfinished round: If you reload during a round, the captured stake remains deducted. The round ends without refund, payout, or continuation.
- Persistence: Settled balance and fractional carry are stored locally. Default stake (20) and first preset are restored after reload.
Settlement and Auto-Banking
Settlement credits the floor of (raw gross return + stored fractional carry). Any remainder persists as fractional carry. Zero-return rounds leave fractional carry unchanged.
- Auto-bank: Triggers automatically only when no eligible uncertain choice remains.
- Disable rule: Choices are disabled if gross > 1000×stake; equality is allowed.
- Collection: You may collect only after at least one safe reveal; you can then collect or choose another eligible cell.
Controls and Modes
Stakes: Selectable at 10, 20, 50, or 100 play coins (default 20). Controls are locked while a round is active.
Hazards: Choose 3, 4, or 5 hazards before the round. Printed labels (1–16) are not clues to a hidden independent grid; they are used only for subset filtering.
Math Evidence: Expected RTP is 97% across all modes (3, 4, 5 hazards). Measured RTPs vary slightly due to sampling error (e.g., 96.949% for 3 hazards, low-1 policy). These are supporting observations, not promised session results.
Privacy and Storage
The demo stores virtual-coin balance and fractional payout carry in browser localStorage. Clearing cached files may leave local storage if not explicitly cleared. The page template loads Google Fonts stylesheets and font files. No account or payment controls are present in the demo.
Audience and Responsibility
This guide and demo have an 18+ intended audience. It is not a universal legal age or implemented age gate. Responsible use includes optional pauses and setting personal virtual-coin limits. The demo does not clinically train behavior or make losses safe. No self-exclusion tools or account-level blocking are provided in this browser demo.
Frequently asked questions
What happens if I reload the page during a round?
The captured stake remains deducted. The unfinished round ends without refund, payout, continuation, or resume. Default stake (20) and first preset are restored.
How is the fractional carry handled?
Fractional carry persists in browser localStorage across settlement, refill, and reload. It is added to the raw gross return at settlement, and any remainder persists.
Is the 97% RTP guaranteed for every session?
No. The expected RTP is 97% across all modes, but measured RTPs vary (e.g., 96.949% to 97.161% in samples). Simulation measurements are supporting observations, not a promised return.
