Every number on this site can be traced to a formula, a source and a date. Here's how.
1. Separate the model, the table and the evidence
A probability model describes how outcomes are generated. A payout table assigns a return to each outcome. Replay evidence tests whether a supported algorithm reproduces captured disclosed bets. These are separate checks. Neither a high theoretical RTP nor a matched replay is an overall casino endorsement.
2. Probabilities and return
Our balanced Plinko model assumes independent left/right steps with probability one half. Under that model, slot k on an n-row board has probability:
P(k) = C(n, k) / 2^nDirectional-bias modes or different game rules require their own probabilities and are outside this balanced model. Risk labels change the captured payout table; similarly named risks at different operators need not have the same payouts.
RTP = Σ P(k) × multiplier(k)House edge is one minus RTP. For independent drops with a constant event probability p, the chance of at least one hit in N drops is 1 minus (1 minus p) to the power N. These are formula-based results subject to the stated model and floating-point rounding. A bankroll estimate or simulation is not an exact guarantee about a future session.
3. Captured operator tables
Each published table identifies its source, capture or check date and version. We record the configurations actually captured, rather than claiming to have every mode, limit or historical table. Sources can include operator documentation, public code and game-interface captures. Operators may change their games after a capture. Use the version that applies to your bet.
4. Replay status
- Verified: within the documented scope, the adapter matched captured disclosed rounds and their saved seed commitments. Evidence notes identify the captures and test vectors. This label describes the adapter, not an operator licence or overall safety.
- Reference: the implementation follows the documented scheme, but does not yet have the same captured-round evidence. Historical community examples and constructed test cases are labelled as such. A Reference verifier can still be available.
- Scope: read the supported mode, algorithm version, nonce format and payout-table limits on the individual verifier. A table snapshot alone cannot verify the result-generation algorithm.
A replay checks the revealed server seed against a recorded hash, reproduces the outcome and, where supported, looks up its payout. A hash calculated only after a bet is not independently recorded evidence of a commitment before that bet. The engine repository documents the available evidence and tests.
5. Simulations and demos
The simulator uses a seeded statistical random generator with the selected probability model and payout table. It runs fixed stakes and stops a session when the remaining balance cannot fund the next stake. Simulations estimate a distribution of possible sessions; they do not reproduce an operator round or predict a future bet. The displayed seed, configuration and version information support repeatable comparisons.
The demo uses virtual credits and its own local result generator. Choosing an operator loads its captured payout table; it does not switch the demo to that operator’s production algorithm. A simulated session can finish ahead even when the model has a house edge.
6. Comparisons and outbound links
Comparisons show theoretical return and the maximum multiplier for the selected captured configuration. Game settings retain each operator’s original risk names. The list has no overall casino score and does not assess withdrawals, solvency or licensing. Affiliate agreements do not change the calculations or list order. Some contextual tool links describe the highest checked RTP among eligible captured configurations; that is not a best-casino recommendation.
7. Claim checks and corrections
The claim-check section explains the planned evidence and verdict format. An empty section is not a list of completed checks. A published check must distinguish its source claim, reproduced result and limitations. Send errors or new source material to hey@provablyhero.com. See the editorial policy for corrections.
8. Limits
We do not audit operators’ servers, balances, payment processing or legal status. A replay cannot prove that an operator used the same process for every bet. A table capture cannot guarantee future payouts or availability. No result here guarantees that a session will be profitable.