How NDB TrustGrade rates online casinos
NDB TrustGrade is a transparent, math-driven score from 0 to 5 that we assign to every casino we list. It blends eight weighted pillars — reputation, player sentiment, bonus value, games, payments, support, popularity, and an editorial AI signal — into a single number. No operator can buy a better score. No affiliate revenue or click data feeds the formula.
What is NDB TrustGrade?
TrustGrade is the headline rating you see next to "Our Review" on every casino and bonus page on this site. It is computed automatically from data we collect (license, year, games, payments, software, traffic, user reviews) plus our internal editorial assessment. We disclose every input, every weight, and every penalty so readers and operators can understand exactly how the number is produced.
How is it calculated?
Each casino gets a 0..1 sub-score on each of eight pillars. The pillars are combined as a weighted sum, multiplied by penalty multipliers (e.g., no license declared, casino closed), then scaled to a 0..5 grade. Casinos with too little verified data are explicitly marked "Not yet rated" instead of getting a fabricated number.
Editorial principles — what does NOT influence the score
A trustworthy rating must be unbuyable. The following inputs are deliberately excluded from the TrustGrade formula:
- Affiliate revenue, commission tiers, or operator payouts have ZERO weight in the score.
- Operators cannot pay to alter, raise, or remove their TrustGrade.
- Internal promotional positioning (featured slots, homepage spots) is excluded.
- Click-through rate, traffic-driven revenue, and conversions are excluded.
How TrustGrade is built — the eight pillars
Each pillar measures one dimension of casino quality. Their relative weights determine how much each dimension can move the final score.
- Reputation & Trust 25.0%
- Player Sentiment 20.0%
- Bonus Value 15.0%
- Games & Content 12.0%
- Payments & Speed 12.0%
- Support & UX 8.0%
- Popularity & Engagement 5.0%
- AI Staff Signal 3.0%
Disclosed regulator (MGA, UKGC, Curaçao, etc.), brand age, closure status, regulator-action history.
Hold a top-tier license, keep operating history clean, stay open and reachable.
Bayesian-smoothed mean of verified user reviews collected on this site.
Earn genuine, positive reviews from real players. We dampen low-volume swings on purpose.
Active offers, wagering requirements, max cashout, exclusivity, no-deposit availability.
Run fair bonuses with reasonable wagering and a meaningful max cashout.
Number of software providers integrated, declared RTP / payout percentage, game variety.
Integrate more reputable providers; publish honest payout statistics.
Deposit and withdrawal methods, crypto support, payout percentage, processing times.
Offer multiple cashout options, including crypto, and pay players quickly.
Mobile readiness, balance of pros and cons in our editorial assessment.
Provide live chat, fast email replies, and a usable mobile experience.
First-party page traffic over the last 90 days, log-ratio normalized to the 95th percentile.
Build a brand players actively look up — we measure attention, not paid clicks.
Editorial-approved staff score, used as a small smoothing prior alongside the other pillars.
Operators cannot tune this signal; it reflects an internal editorial review.
The math (for the curious)
Click any section to expand. All formulas are real, runnable, and identical to what the codebase computes — no marketing approximations.
Final formula
where w_i is the configured weight for pillar i (sum equals 1.0), s_i is its sub-score in [0, 1], and p_j are penalty multipliers in [0, 1] applied for the conditions listed in the next section.
Player Sentiment — Bayesian smoothing
- $R$ — average user rating for this casino (5-point scale, sourced from on-site reviews).
- $v$ — verified review count.
- $C$ — global cross-site review average (re-computed every 6 hours).
- $m$ — prior weight (default 10), tuned to dampen low-volume swings.
Popularity & Engagement — log-ratio normalization
Where v_i is the page-views count over the last 90 days for casino i, and v_p95 is the 95th-percentile of view counts across all casinos in the same window. We use first-party tracking via Post Views Counter Pro — no third-party trackers feed this signal.
Publication gate — when we refuse to publish a number
A casino must have verified data on at least 4 of the 8 pillars, AND have data on either Reputation OR Player Sentiment, before we publish a numeric TrustGrade. Otherwise we display "Not yet rated" and exclude the casino from review JSON-LD.
How does a casino get a better TrustGrade?
Operators sometimes ask us how to improve their score. The honest answer: do the work, on every pillar. Below are concrete, non-cosmetic improvements that move the number.
Penalty multipliers
Penalties are applied multiplicatively. Multiple triggers compound, which is why a closed, unlicensed casino can drop dramatically below its raw weighted sum.
closed× 0.20Casino is editorially flagged as closed or no longer operating.
no_license× 0.70No regulator declared in our records.
low_tier_license× 0.92Lower-tier offshore jurisdiction declared.
new_brand× 0.92Brand established less than one year ago.
low_data× 0.95Fewer than three pillars had real data available.
Update cadence
Why can readers trust NDB TrustGrade?
Every input, weight, formula, and penalty is documented on this page. The score is generated by code, not by a human deciding what a casino "should" rate. Editorial overrides exist for exceptional cases (e.g., regulator action, confirmed scam reports), but they are stored with a reason, an auditing user, and a timestamp, and are clearly labeled on the casino page.
Algorithm changelog
- v1.0.0 First public release: 8 pillars (Reputation 25%, Sentiment 20%, Bonus 15%, Games 12%, Payments 12%, Support 8%, Popularity 5%, AI Staff 3%); penalty multipliers for closed, no_license, low_tier_license, new_brand, low_data; publication gate at ≥4 available pillars with at least one anchor pillar.