# CasinoGuard Methodology **How CasinoGuard collects, preserves, evaluates, and publishes evidence-backed casino data.** CasinoGuard maintains an immutable, evidence-backed canonical dataset of the online casino industry. This methodology defines how source information is collected, how observations are preserved, how claims are evaluated, and how findings are published for independent examination. The methodology is designed around five principles: 1. **Evidence precedes claims.** Findings must be traceable to identifiable observations or sources. 2. **Sources are not interchangeable.** A casino's statement, a regulator's record, and a direct observation establish different things. 3. **Historical evidence is preserved.** New information does not silently replace previously captured observations. 4. **Uncertainty is explicit.** Missing, conflicting, and unconfirmed information is not presented as established fact. 5. **Verification is independent of commercial influence.** Findings are determined by documented evidentiary rules rather than promotional or editorial preferences. ## 1. Data collection CasinoGuard collects information through automated workflows that retrieve publicly accessible casino websites, official documents, regulatory records, and other relevant sources. Collection is performed under identifiable observation conditions. Where applicable, the collection record includes: - Source URL and final resolved URL. - Collection date and timestamp. - Geographic location of the request. - Device or browser context. - HTTP response information and redirects. - Captured page content and associated data responses. - Screenshots, archives, and other preserved artifacts. Observations are made from specific geographic locations because casino websites may present different information depending on the visitor's country. Where supported, collection may also account for device-specific or language-specific variations. CasinoGuard does not treat a failed collection attempt as evidence about the casino's underlying characteristics. A blocked page, inaccessible resource, or unsuccessful request does not establish that the information is absent. Automated crawlers do not attempt to circumvent website bot protection. Cookie consent interfaces are handled using the most privacy-preserving available choice. Only successfully retrieved and interpretable source observations are used as substantive evidence. ## 2. Evidence preservation and immutability Evidence preservation is a foundational property of CasinoGuard. Collected observations are stored with persistent identifiers and associated provenance metadata. Depending on the source and collection method, preserved evidence may include: - Screenshots of relevant page sections. - Full-page screenshots. - WACZ web archives. - Original website or API responses. - Source and redirect URLs. - HTTP response details. - Collection timestamps and geographic context. - SHA-256 hashes of evidence artifacts. Evidence records use persistent identifiers beginning with `ev_`. These identifiers allow published claims to reference specific historical observations rather than relying exclusively on the current state of an external website. When source content changes, the new observation is recorded separately. Previous evidence remains part of the historical record. The dataset distinguishes between immutable evidence artifacts and the evolving interpretation of those artifacts. A new observation may justify updating, correcting, or superseding a claim without altering the historical evidence on which an earlier finding was based. Cryptographic hashes support file-integrity verification. They do not, by themselves, constitute independent proof of the completeness or append-only nature of the entire dataset. ## 3. Sources and provenance Every published finding must have an identifiable evidentiary basis. CasinoGuard distinguishes between four principal source categories. ### Regulatory sources Official licensing authority registers, regulator-operated validation services, and other relevant governmental or regulatory records. These are the preferred sources for verifying licensing information and the legal entities identified in the relevant records. A regulatory record is interpreted within its jurisdiction, scope, and observation date. ### Official casino sources Information published directly by the casino, including its website, terms and conditions, cashier, game lobby, promotions, and official data responses. These sources are primary evidence of what the casino publicly presents. They are not automatically independent confirmation that every statement made by the casino is true. ### Direct observations Information measured or observed through CasinoGuard's own collection process. Examples include whether a domain loads from a particular country, what content is displayed to a visitor, and which payment methods appear in an accessible cashier. Direct observations establish what was observed under the recorded conditions. ### Third-party sources Independent publications, review websites, affiliate directories, forums, and other external information providers. These sources may provide useful context or corroboration. However, repeated publication of the same underlying assertion does not create independent verification. Third-party information is not treated as equivalent to direct regulatory confirmation or an applicable primary observation. **Source authority is evaluated relative to the claim being examined. No single source category is universally authoritative for every type of finding.** ## 4. Atomic claims and claim provenance CasinoGuard represents findings as structured claims connected to supporting evidence. An atomic claim describes a specific proposition that can be independently examined. Examples include: - A named company appears as the operator in a licensing authority's register. - A casino displays a particular payment method to visitors from a recorded country. - A promotional offer specifies a wagering requirement in its published terms. - A casino's website identifies a particular licensing authority. - A disclosure is absent from a specific page captured under recorded conditions. Claims must be interpreted within the scope of their evidence. For example, observing that a casino displays Mastercard to visitors from Canada does not establish that Mastercard is available in every country or that every attempted payment will succeed. A statement that rests on several readings has a claim page, identified by `cl_`. A claim page provides the relevant finding, its evidentiary basis, and references to the underlying observations. A `cl_` identifier only groups the evidence for a specific statement. It is generated, not stored: a SHA-256 hash of the sorted identifiers of the claims behind the statement, shortened to 20 characters. Statements resting on the same claims share one `cl_` identifier, and it changes when any of those claims gets a new version. A `cl_` identifier is therefore not a stable reference. The persistent references are the `ev_` identifiers of the evidence records, which remain valid when a claim page changes. Where a claim aggregates multiple readings, those readings remain individually identifiable. Claim identifiers and evidence identifiers are distinct from page-local citation numbers. A citation number such as `[1]` identifies a reference within a particular published page. Persistent `ev_` identifiers allow the underlying evidence to be referenced independently. ## 5. Verification statuses CasinoGuard uses five verification statuses to describe the strength and limitations of the available evidence. Verification status is determined according to the relevant source requirements and the subject being examined. ### Verified The applicable primary source or direct observation confirms the finding within its stated scope. For operator and licence information, this ordinarily requires an applicable licensing authority record or official validation service. For directly observable website characteristics, verification may establish that particular information was displayed under recorded conditions. Verified does not mean that all possible sources have been examined or that every real-world implication of the observation has been independently tested. ### Supported Two or more independent publishers support the same finding, but the required primary-source confirmation is unavailable. Multiple pages controlled by the same casino count as one publisher for independence purposes. Supported indicates corroboration, not definitive primary-source verification. ### Reported A source makes the relevant assertion, but sufficient independent or primary-source confirmation is unavailable. The source of the assertion remains explicit. For example, an operator identity disclosed only on a casino's website may be Reported rather than Verified. ### Disputed Available evidence contains materially conflicting information relevant to the finding. The disagreement is documented, with each position attributed to its respective source. Disputed findings are not silently resolved by selecting the most convenient statement. ### Unknown The available evidence does not establish the requested information, or the relevant information has not yet been collected. Unknown may indicate missing disclosures, insufficient observations, or incomplete coverage. Unknown is not equivalent to a negative finding. A licence that could not be confirmed is not automatically classified as nonexistent. ### Verification and evidence age Verification status describes the strength of evidentiary support, not the age of the observation. ## 6. Verification rules by data category Different kinds of information require different evidence. CasinoGuard applies subject-specific verification rules rather than treating every casino disclosure as equivalent. | Data category | Primary evidence | Verification principle | |---|---|---| | Operator | Licensing authority register | Verify the registered company; casino-only disclosures remain attributed | | Licence | Regulatory register or official validator | Confirm licence details with the relevant authority | | Domains and mirrors | Direct geographic access observation | Verify that the domain was observed to load under recorded conditions | | Game providers | Captured game lobby | Verify providers displayed in the observed location | | Payment methods | Captured cashier | Verify methods displayed in the observed location | | Languages and currencies | Official casino interface or pages | Verify the published or displayed options | | Withdrawal limits | Official terms or relevant account interface | Verify the stated conditions, not actual payout performance | | Bonus conditions | Official offer terms | Verify the conditions published for the captured offer | | Promotions | Individual promotion pages | Verify each offer from its own terms where available | When only a weaker source is available, the finding is attributed and assigned an appropriate verification status. For example, a promotion mentioned on a general promotions page but not independently captured from its own offer page may remain Reported. Where applicable sources disagree about an operator or licence, the conflict is preserved as Disputed. Verification is always limited to what the relevant evidence actually demonstrates. ## 7. Geographic observations CasinoGuard treats geographic context as an essential component of casino data. Casino websites may display different operators, licences, payment methods, promotions, restrictions, and terms depending on the visitor's country. Observations are therefore associated with their collection geography. Country-specific findings describe what was observed from the corresponding location. General or aggregated findings combine the available geographic observations while preserving relevant differences. An aggregated finding is not a universal assertion. If observations exist for Canada, Finland, Italy, and Poland, the resulting aggregation describes those four observed geographies. It does not establish what visitors in unobserved countries would see. Conflicting country-specific disclosures are retained rather than flattened into a single global value. A country without observations is treated as unobserved, not as a country where the feature or disclosure is absent. ## 8. Missing information and negative evidence CasinoGuard distinguishes between information that is absent from an observed source and information that does not exist. These are not equivalent. For example: - An operator not named in a captured footer is recorded as not disclosed in that footer. - A payment method not displayed in a captured cashier is recorded as not observed under those conditions. - A licence not found in a checked register is recorded within the scope of that register search. - A country not examined remains outside the observation coverage. The dataset does not automatically convert these observations into universal negative claims. Statements such as "unlicensed," "unavailable worldwide," or "no withdrawal limits" require evidence sufficient to support those specific conclusions. **Absence from an observation is not automatically proof of universal absence.** ## 9. Conflicting evidence When sources disagree, CasinoGuard preserves the disagreement and identifies its origin. For example, a casino website may name one operating company while a regulatory register identifies another. The dataset records both statements, their sources, the observation dates, and the applicable verification status. The stronger source may establish a more authoritative finding for a particular question, but weaker contradictory evidence is not silently removed. Where different geographic observations produce different disclosures, those differences remain attached to their respective locations. Conflicts may result from outdated information, jurisdiction-specific arrangements, changes in ownership, or inaccurate disclosures. CasinoGuard does not assign an explanation without supporting evidence. ## 10. Historical changes and reassessment Casino information changes over time. Operators may change, licences may expire, payment methods may be added or removed, and promotional terms may be revised. CasinoGuard preserves earlier observations while allowing current findings to be reassessed as new evidence becomes available. A newly captured source does not erase an earlier source. Historical evidence remains identifiable through its persistent record. Claims may be updated, corrected, or superseded to reflect the strongest currently available evidence. Where a historical finding differs from a current finding, the relevant dates and evidence should allow the distinction to be understood. The objective is to maintain an auditable history rather than an endlessly overwritten description. ## 11. Automated processing and error control CasinoGuard uses automated collection and evaluation workflows to reduce dependence on subjective editorial judgment. Automation supports consistent data collection, source identification, evidence preservation, comparison, and structured publication. However, automated systems can encounter incomplete pages, misleading source material, parsing errors, geographic inconsistencies, or incorrect interpretations. CasinoGuard does not treat automated output as inherently authoritative. The methodology requires published findings to remain traceable to their underlying evidence. Where evidence is insufficient or ambiguous, the finding must reflect that limitation rather than introducing unsupported certainty. Corrections and reassessments should be based on new evidence or identifiable errors, not on commercial requests or editorial preference. ## 12. Commercial independence CasinoGuard is not an affiliate ranking or promotional publishing platform. Its verification methodology is independent of casino advertising, operator preferences, and affiliate relationships. Commercial parties cannot purchase a favourable verification status, require the removal of unfavourable evidence, or override the dataset's evidentiary rules. Any source, including a casino operator, may provide information relevant to a finding. Such information is evaluated according to its provenance and evidentiary strength. The identity of the submitting party does not independently establish the truth of the information. ## 13. Machine-readable publication and citations CasinoGuard publishes structured information for both human examination and machine retrieval. The dataset includes canonical casino pages, country-specific records, claim pages, evidence pages, and machine-readable representations. Available interfaces include: - Human-readable casino records. - Markdown representations using the `.md` extension. - JSON representations using the `.json` extension. - Versioned API access through `/api/v1/casinos/{slug}`. - Geographic queries such as `?geo=FI`. - Evidence records identified by persistent `ev_` identifiers. - Claim pages identified by `cl_` identifiers, which group the evidence behind a statement. - Machine-discovery resources including `llms.txt` and an XML sitemap. Published summaries are intended to remain understandable when retrieved independently. Citations connect findings to the observations and evidence used to establish them. AI systems consuming CasinoGuard data should preserve the distinction between a verified observation, a reported assertion, and an unresolved conflict. They should also retain relevant dates and geographic limitations when presenting retrieved information. ## 14. Methodological limitations CasinoGuard documents what the available evidence supports. The dataset cannot guarantee that every source is truthful, that every relevant jurisdiction has been examined, or that every published observation remains current. A website capture proves what was displayed under recorded conditions, not necessarily the accuracy of the website's statements. A regulator's record establishes information within the authority's published scope, not a guarantee of future operator behaviour. Cryptographic hashes support evidence integrity but do not independently establish the completeness of the entire collection. Automated analysis can produce errors, and downstream AI systems can misinterpret otherwise accurate data. CasinoGuard therefore prioritises explicit uncertainty, preserved evidence, transparent provenance, and the ability to reassess findings. The methodology is intended to make errors discoverable and findings independently examinable, not to claim infallibility. ## 15. Methodological commitment CasinoGuard's methodology is designed to keep evidence separate from assertion, verification separate from repetition, and historical observations separate from current interpretation. Published findings must remain grounded in identifiable evidence. Conflicting information must not be silently concealed. Missing information must not be replaced with assumptions. Commercial interests must not determine verification outcomes. Historical evidence must remain preserved when knowledge changes. **Evidence is preserved. Claims are evaluated. Nothing is accepted on authority alone.** For the project's purpose, independence, and architecture, see [About CasinoGuard](https://casinoguard.ai/about).