Open data

Every number on this site comes from a simulation we ran or a document we read. Both are published here as plain files. No signup, no email, no API key. Reuse them with attribution, and if you find an error we would rather hear about it than not — the correction goes on the page, dated.

Simulation code and raw results

Standalone Python 3, no dependencies beyond the standard library, seed in each docstring. Run one and you get the figures we published, or you find a bug in our work. Either outcome is more useful than taking our word for it. The studies themselves are indexed in our research index.

An Aviator-variant crash result set is also published: crash-sim-aviator-results.txt.

Measured house edge

Every major game and bet ranked by house edge, with the cost per $100 wagered and, for each, whether the figure was measured in our own simulations, derived in closed form, or taken from a published RTP we did not independently check. The distinction is the point: most tables on the internet mix all three and label none.

house-edge.json 15 games and bets, 0.44% to 16.67%, with method and sample size

Generated by parsing the ranked table in the guide itself, so the data and the article cannot disagree. If the parse ever stops producing a plausible result the endpoint returns the failure rather than a partial dataset.

Operator restrictions and promotion law

Two datasets that, as far as we can tell, nobody else publishes with sources attached. The first is each operator’s own restricted-country clause, read at source and archived. The second is our record of the countries where advertising an operator that is not locally licensed is an offence for the publisher, not merely where playing is an offence for the player — with the statute and how far we verified it.

operator-restrictions.json per-operator refusal lists, promotion-law rows with statutes, programme status

The human-readable version is the country acceptance checker, and the reasoning behind the promotion-law layer is set out in legality by country.

VIP ladder thresholds

The two crypto casinos that publish their VIP tier thresholds, as data: every tier with the points required, the earning rate per dollar wagered, and the source. Each ladder carries its own grade, because one was read on the operator’s own pages and the other could not be — stake.com answers HTTP 451 from where this site is maintained, so its 15 tiers are reconstructed by cross-checking two independent sources that agree.

Per-tier rakeback percentages and level-up bonus amounts are absent because neither operator publishes them. Any dataset that includes them is guessing.

vip-tiers.json Stake 15 tiers, Shuffle 54 tiers, earning rates, sources and verification grade

The interactive versions price the climb rather than just listing it: Stake VIP calculator and Shuffle VIP calculator.

Brazil’s authorised operator register

Every company the Secretaria de Prêmios e Apostas authorises to offer fixed-odds betting in Brazil, with the ordinance that authorised it, its CNPJ, and each of its brands and .bet.br domains. The government publishes this as a semicolon-delimited CSV three levels down a ministry site, laid out with merged cells so that reading it row by row gives 188 nameless companies instead of the 85 real ones.

brazil-spa-register.json 85 companies, 188 brands, with CNPJ, ordinance and authorised domain

It matters because the register is the only thing that decides legality: under Portaria SPA/MF 1.231/2024 an affiliate link may point only at an operator listed on it. Reading it corrected two of our own reviews, which carried domains that do not resolve. The narrative version is every betting site legally authorised in Brazil; the lookup is the authorisation checker, and the authorised operators we have reviewed are ranked in best authorised betting sites in Brazil.

Curaçao’s full licence register

Every row of the Curaçao Gaming Authority’s Online Gaming License Registry, which the regulator publishes only as a dated PDF: 665 licences in the edition of 21 September 2026, with licence number, licensee, B2C or B2B, Chamber of Commerce number, dates and status exactly as printed, plus the 32 entries of the separate enforcement register. Tabulated, 281 licences read “Assessment in progress” and 276 of those are past the expiry date printed in their own row; all 50 revocations in the licence register were at the operator’s request.

curacao-cga-register.json curacao-cga-register.csv 665 licences, enforcement register, summary counts; regenerated on the 45-day cycle

The findings, with the method and its guard against partial reads, are in Curaçao’s licence register as open data.

Licence register readings

The rows we copy from the Curaçao Gaming Authority’s Online Gaming Licence Registry PDF, verbatim, on a 45-day cycle. Status labels are the registry’s own. Anjouan licensees are absent because Anjouan publishes no equivalent register, which is itself the finding.

licence-register.json entity, licence number, issue and expiry dates, status, registry date

The dated narrative version, with a change log, is the licence tracker.

How to cite this

Attribution to casinoaurum.com with a link, plus the date of the reading where one is given. The licence-register and operator-restriction datasets carry their own read dates inside the file, because a clause or a registry row is only true as of the day it was read.

18+ only. These datasets describe gambling products. Gambling involves real financial risk, and nothing here is a recommendation to gamble.