How we calculate TrustScore: the full method, with the numbers

A casino with three five-star reviews is not better than a casino with three hundred reviews averaging 4.4. A plain average would say it is. TrustScore is our attempt to produce a single number that behaves sensibly in both cases, and this guide explains it in enough detail that you could reproduce every score on the site. The live constants and worked examples are always on the How we rate page, rendered from the same code that scores casinos.

Step 1: only published reviews count

A review enters the calculation only when its status is published. Reviews held for moderation (a first review from a new account, a review containing a link, a review that tripped our screening rules) do not count until a human approves them. Rejected reviews never count. Reports from other users never change a review's status by themselves — they queue it for a human — so a campaign of reports cannot remove a review from the score.

Step 2: weight each review by how recent it is

Each published review gets a weight based on its age. A review published today has weight 1.0. The weight halves every 365 days (the half-life), so a one-year-old review counts 0.5, a two-year-old review 0.25, and so on. The weight never drops below 0.1, so even very old reviews retain a small voice.

The formula: weight = max(0.1, 0.5 ^ (ageInDays / 365)). This means a casino that fixed its withdrawal problems eighteen months ago can recover its score as new reviews arrive, while a casino that has recently gone bad shows it quickly.

Step 3: shrink toward a neutral prior

We pretend every casino starts with seven imaginary reviews at 3.5 stars. The score is then the weighted average of those seven phantom reviews plus the real ones: score = (7 × 3.5 + Σ weight × rating) / (7 + Σ weight).

With few real reviews the phantom reviews dominate and the score sits near 3.5. As real reviews accumulate their weight overwhelms the prior and the score converges on the true weighted average. Three fresh five-star reviews give (24.5 + 15) / 10 = 3.95, displayed as 4.0. Five give 4.1. Forty give about 4.78, displayed as 4.8. No small number of reviews can produce a 5.0; we consider that a feature, and we say so on every profile with an Early rating label.

• Prior size: 7 reviews. Prior mean: 3.5 stars.
• Half-life: 365 days. Minimum weight: 0.1.
• Result clamped to 1.0–5.0 and rounded half-up to one decimal.
• Zero published reviews: score 0, shown as "No reviews yet" with grey stars.

Step 4: display rules

The star widget shows the score rounded to the nearest half star — 4.3 shows four and a half stars, 4.2 shows four — following the convention most review sites use. Every filled star takes the colour of the score's band: Excellent (4.3 and above) green, Great (3.8–4.29) light green, Average (2.8–3.79) yellow, Poor (1.8–2.79) orange, Bad (below 1.8) red. Individual reviews show their own integer rating in the colour of that rating's band.

A casino with one to four published reviews shows its score but carries an "Early rating" label, and it is excluded from top-rated lists on the homepage and category pages until it reaches five reviews. The rating distribution bars show raw counts and percentages of published reviews, unweighted, so you can see the shape of opinion behind the number.

What never affects the score

TrustScore uses only published player reviews. Bonus size and minimum deposit enter neither TrustScore nor the separate facts score; payout and payment methods are inputs only to the facts score. A licence claim and a Verified review badge do not enter TrustScore. Commercial partners are labelled. Before enough player reviews exist, the default numbered list uses a disclosed editorial order, with TTBet and BCB88 first and second at the owner's request; that order is not a player rating.

The "Most relevant" ranking on listing pages multiplies TrustScore by a confidence factor that grows with review count and reaches full weight at 25 reviews, so a 4.6 with 40 reviews outranks a 4.7 with 6. "Highest rated" sorts by raw score. Both are documented on the How we rate page.

Known limitations

The score reflects the players who chose to write reviews, who skew toward people with strong experiences in either direction. Our screening catches obvious manipulation — duplicate bodies, links and promo codes, brand-new accounts — but it cannot detect a patient operator seeding plausible positive reviews over months. That is why we show review counts, Early rating labels, distribution bars and issue tags alongside the number, and why we ask you to read the recent low-star reviews rather than trusting a single figure. On this demo deployment, all casinos and reviews are fictional seed data; the method is real.