The ranking rule, written out in full

Every check has two possible answers, pass or fail, and a pass earns a fixed number of points. The ten weights add up to 10, which is also the most any brand can reach. The final figure is rounded to one decimal place.

Two checks carry 1.5 points rather than 1: whether the record holds a licence status that is not flagged as revoked, and whether the operator's terms describe identity verification at all. They got the extra half point because a reader is most exposed when either of those facts is missing.

A brand that publishes two facts is scored out of 10, exactly like a brand that publishes all ten. The tempting alternative divides by however many fields happen to be filled in. Under that version an operator could publish one correct fact and collect a perfect mark, which pays brands for saying as little as possible.

With the maximum held still, a low score is simply an accurate description. A brand on 2 out of 10 has not been judged a bad casino. It is a casino for which eight of the ten questions had no published answer, and the number reports exactly that.

On this site, a blank field is always a sign that the source was missing or unread, and never a way of saying the answer is nothing. Code that mixed the two up would hand out points for documents nobody opened, so the scoring treats an empty list as a failed check.

Before any scores are compared, the sort splits the records into two groups. An operator whose licence status reads as revoked, withdrawn or cancelled goes into the lower group, and every operator in the upper group ranks above it, however the points fall.

In the current records that lower group holds one brand, Bets.io, whose licence the register returns as revoked. Its other fields may add up to a respectable total. That changes nothing here: no amount of published detail can lift a brand over an authority that has taken its licence away.

Within each group, higher scores come first, and brands on equal scores are listed alphabetically so the table does not reshuffle from one build to the next. Apart from that tie-break, the revoked-licence rule is the only step where something other than arithmetic sets a position.

The score asks whether an operator published a restricted-country list at all. Which countries the list names plays no part in any score. Publishing the list is treated like publishing a coin list: a sign that the operator is willing to put its rules in writing.

The contents are read in one place only, the data build that decides who appears on this site. A brand is included when its own list was read and Japan is not on it. Because every listed brand has passed that filter, every listed brand also passes this check, so in practice it adds the same 0.75 to each row and separates nobody.

The two jobs are kept apart on purpose. If the score read the contents of restricted lists, it would start rewarding or penalising brands for where they choose to trade, and this site ranks only what operators publish.

The score cannot see how quickly withdrawals arrive, how support answers, how disputes end, how generous a bonus is, or whether an operator keeps to its own clauses. Measuring any of that would mean playing with real money, and nothing on this site involved that.

Size and fame earn nothing either. There are no points for traffic, popularity or age. Stating a founding year earns half a point on its own, so a brand that opened in 2013 and one that opened in 2024 receive the same half point if both give the year.

Money sits outside the routine too. The operator record has no field for a commercial deal, so a payment has no way into the score. The site's paid partner, Vave, is not a row in the ranked table; on the front page it has its own block, labelled as a commercial placement.