How the sheet works
We review public operator pages, terms, fairness documentation and dated third-party reports. No deposit, withdrawal or KYC test has been performed for the current catalogue.
The catalogue covers affiliate partners, so it is not a survey of every available operator. Each score totals five fields out of 100. Affiliate payments receive no points; the score measures the public documentation we found, not a tested gambling experience.
| Field | Max | What it measures |
|---|---|---|
| Documentation | 30 | Named operator, address, terms, fairness page, cashier rules in public. Missing legal identity is a hard cut. |
| Fairness docs | 20 | A verifier you can actually run (commit–reveal, seeds, published test vectors). A “provably fair” label with no inputs scores low. Fairness is not safety and not RTP. |
| Cashier / spread | 20 | Whether deposit and withdraw paths, fees, and item valuation are described. Skin sites that hide the withdraw price lose here. |
| Product fit | 15 | Does the public lobby match the category the row sits in. |
| Public track | 15 | Age of the brand, named complaints history, third-party review volume. Not Trustpilot stars taken at face value. |
What a high score is not
A high score means more of our documentation criteria were met. It does not establish safety, reliable payouts or positive expected value. Gambling games carry a house edge.
Criterion values involve editorial judgment; they are not an independent audit or a mechanically reproducible measurement. The publisher-requested Rust rating adjustment on 3 October 2026 did not add new evidence. See the rating change log for the previous values and rationale.
Evidence labels
- High — named entity or unusually complete public fairness/cashier docs.
- Medium — brand is real, several fields still open.
- Low — too little verified information for a recommendation. The listing remains incomplete.
Algorithm checks are not transaction tests
A public synthetic fixture can be recomputed without registering, depositing, wagering, completing KYC, or withdrawing. We label that result “algorithm reproduced.” A profile becomes “transaction tested” only when it has a real dated test record with redacted evidence.
Rust Upgrader Transparency Index
The separate 100-point index uses fixed criteria for published RTP, exact chance calculation, pre-play commitments, round verification, downloadable fixtures, custody rules, operator identity, and evidence freshness. Affiliate status receives zero points. The public JSON and CSV contain the same values shown on the page.
Refresh
Bonuses go stale in days. Structural fields (entity, licence, game list) should be re-read monthly. Anything labelled unresolved stays unresolved until a dated source fills it — we do not invent percentages.