Atlas methodology
How Atlas scores bookmakers
The Atlas Trust Score is a deterministic 0 to 100 number built from six weighted components, capped by red-flag overrides for severe signals (sanctions matches, third-party trust warnings). Scores recompute monthly from raw data; no editorial nudging happens between runs. Atlas ignores affiliate-commercial relationships entirely.
1,869 brands scored76 regulators tracked6 weighted components50+ public data sources
Last recompute: 2026-06-10
The formula on a live brand
89/100
- Licensing
- 100 ×30%
- Corporate
- 82 ×20%
- Security
- 60 ×10%
- Reputation
- 82 ×15%
- Threat intel
- 90 ×10%
- Operational
- 100 ×15%
Weighted sum 88.7 rounds to the published 89. No red-flag override applied. The same math runs on every brand page, recomputed monthly.
The 6-component Trust Score
Each component scores 0 to 100 from raw signals, then multiplies by its weight. Components without data fall back to a neutral 50 with weight halved, so missing data never silently boosts a score.
How the 100 points split
When a component cannot be measured for a brand, it contributes a conservative neutral prior at half weight; unknown is neither free nor fatal. On top of that, coverage caps apply: a brand verified on only 3-4 of the 6 components cannot score above 75, and 2 or fewer cannot score above 60, no matter how clean the few measurable signals are. Top-band trust requires comprehensive verification.
Licensing
30%How strong the strongest currently-valid licence is, plus licence breadth. A licence counts as active whenever the regulator lists it as live (Licensed, Authorized, Active and local equivalents), not only when it is literally tagged "Active". Top-tier regulators score highest (the strictest markets, including the major US state regulators alongside the UK, Malta, the Nordics and Ontario); mid-strength score mid; light-touch offshore licences score low; no live licence scores zero and caps the overall score below the top band, because a brand with no verifiable licence offers no consumer-protection recourse no matter how strong its other signals are. Holding many active licences across regulators earns a breadth bonus: one licence and thirty are not the same trust posture.
Raises: an active top-tier licence verified on the regulator register; many active licences across regulators
Lowers: lapsed, surrendered, or only light-touch licences
Corporate transparency
20%Whether the legal entity behind the brand is identifiable. Disclosed parent company, verifiable corporate registration, and a traceable ownership chain raise the score. Hidden or dead-end ownership lowers it.
Raises: a disclosed listed parent, a verified corporate registration, deep ownership chain to a known group
Lowers: no licensee company name, no parent group disclosed, dead-end ownership chains
Security
10%Automated third-party technical scans, not our editorial judgement: web-security grades, transport-layer protection, DNS hygiene, and infrastructure resilience. Signals are matched on the resolved hosting, which can be shared cloud or CDN infrastructure, so we treat this as a supporting signal and never publish specific findings on a brand page. A clean, modern stack scores high; an exposed legacy stack scores low.
Raises: top web-security grades, full DNS hygiene, mature infrastructure, no critical vulnerabilities
Lowers: failing security headers, missing email-auth records, multiple critical vulnerabilities on exposed infrastructure
Reputation
15%Aggregate public reputation signals weighted by sample size, plus app-store ratings on verified store listings. Long-tail trend deltas are tracked across multi-year horizons rather than single-point ratings.
Raises: consistently positive aggregate reputation at high sample size, improving multi-year trend
Lowers: consistently negative aggregate reputation, deteriorating multi-year trend, negative community sentiment
Threat intelligence
10%Whether the brand or its resolved infrastructure appears on independent, automated third-party threat feeds: phishing, malware, sanctions, anti-spam zones, actively-exploited vulnerabilities (CISA Known Exploited Vulnerabilities), takedown-notice volume, and parsed regulator enforcement actions.
Raises: no hits on any threat feed, no enforcement on the licensee company
Lowers: phishing or malware flag, third-party trust-warning, sanctions match, an actively-exploited vulnerability on the resolved stack, recent regulator fines
Operational
15%Brand longevity and reach: domain age, real-world popularity rank, hosting tier, breach exposure, suspended-brand flags, accepted-country count, Wikipedia presence, and archived activity over time. This is the establishment axis that separates household-name operators from thin-data unknowns.
Raises: 10+ year domain age, top popularity rank, broad market footprint, mainstream-tier hosting
Lowers: sub-1-year domain, no archived activity in 12 months, hosted on flagged infrastructure
Red-flag overrides
The weighted average is the starting point, not the verdict. Severe signals cap the final score so a brand with strong licensing or marketing cannot wash out a critical safety flag. Each override applies independently; the lowest cap wins.
- A severe third-party trust-warning caps the overall score
- A sanctions-list exact match caps the overall score
- A regulator licence revocation in the last 24 months caps the score
- A suspended-brand flag in our records caps the score
- A high-confidence anti-spam zone listing caps the score
- Caps are conservative on purpose. A brand can recover from a cap by remediating the underlying signal (getting off a threat-intelligence feed, settling an enforcement action, addressing a sanctions hit). The next monthly recompute will reflect the change.
Data sources
Atlas Trust Score is fed by six categories of public-source data, each on its own refresh cadence. None of the inputs are AI-generated and none are manually overridden after ingest.
| Category | Refresh cadence | What it measures |
|---|---|---|
| Licensing and enforcement | Weekly to quarterly | Active licences across every market we track, regulator strength, recent fines, suspensions, and revocations. |
| Corporate ownership | Quarterly | Parent group, ultimate-owner disclosure, sanctions exposure, and the ownership chain from licensee back to the controlling entity. |
| Web security and infrastructure | Monthly to quarterly | Independent web-security grades, transport-layer posture, DNS hygiene, hosting tier, certificate-transparency activity, and exposure to known vulnerabilities. |
| Public reputation | Monthly | Aggregate sentiment weighted by sample size, multi-year trend deltas, and complaint outcomes across the public-reputation surface. |
| Threat intelligence | Weekly | Phishing, malware, sanctions, anti-spam zones, actively-exploited vulnerability cross-references (CISA KEV), and takedown-notice volume. |
| Brand activity and reach | Quarterly | Domain age, archived activity over time, breach exposure, accepted-country count, and infrastructure classification. |
Calibration sample
After every full scoring run, we spot-check around 20 well-known brands across the trust spectrum to confirm the score lines up with reality. Brands with the strongest regulators usually land in the 70 to 90 range. Mid-strength regional brands sit between 40 and 70. Offshore or sanctioned brands score below 30. If several brands score where intuition says they should not, we revisit the weights and run the score again.
The full calibration write-up is archived alongside the methodology. Specific calibration questions go through our contact page.
Limitations
- Atlas Trust Score is automated. It is not a manual audit, not a substitute for legal advice, and not a guarantee of withdrawal experience.
- Some signals reflect public perception, which can lag reality on both sides. We counter-balance with hard-data signals where possible.
- Data lag exists. A licence revoked yesterday may not appear until the next monthly regulator-register refresh.
- Country-level access is self-declared by operators or read from their published terms. We cross-check where regulator data exists, but accuracy varies by jurisdiction.
- Licensed sister-brands sharing infrastructure with a parent licence inherit the licensing-tier signal; the corporate component captures the relationship separately.
- Last updated: the most recent Atlas Trust Score recompute completed on 2026-06-10. The full leaf set is rebuilt in a single deterministic pass.
Recent data changes
Detected changes from monthly refreshes of score, confidence, or licence counts.
- 2026-06-105 changesbetika, bwin, caesars +2 more
- 2026-06-101774 changes711, 6686, 1731788 +1771 more
Conflict-of-interest disclosure: Sharkbetting earns affiliate revenue elsewhere on the site (oddsmatcher signups, exchange referrals), but Atlas pages have no affiliate kickback and no operator pays for placement here. The Atlas Trust Score ignores commercial relationships entirely. The same scoring code runs against every brand, and component weights are not adjusted per brand. The full editorial framework (independence, corrections policy, AI usage, contact) lives on the Editorial Policy page.
Browse the regulator hub, the best bookmaker lists, or any individual brand profile to see Atlas Trust Score in action.