A best requires a measure, and here is the one we can compute

This is where the withdrawal cap of 100,000 USD per week in clause 9.6 comes from. We found the permit number OGL/2024/210/0198 in the Curaçao certificate register, and the company on the permit is Stack Gaming Ltd.

The measure we actually compute

The bar in the header is eight checks of what the operator publishes, summed to ten points. The checks are these:

1. A permit number is written out. Counts twice.
2. The number can be found at the issuer rather than only on the operator’s own site.
3. A company name is written out on the permit.
4. An amount for the identity check is published. Counts twice.
5. The clause number for that threshold is given.
6. A cap on withdrawals is published.
7. A list of coins accepted is published.
8. The text could be captured directly rather than from an archived copy.

The two that count twice do so for the same reason: they are the only two items that tie the operator to something checkable. A permit number can be looked up, and a published threshold is something an operator can be held to.

What the ten score

Rocketpot sits highest on ten points, Bitcasino.io has nine and Wild Fortune eight. Six operators sit on seven points each: DuckDice, Empire.io, Metaspins, Rainbet, Wild.io and Wolf.bet. The tenth row, the advertising partner Vave, is left without a bar: its documents were read after the count on 26 August 2026 and await rescoring.

That six operators sit on the same figure is itself a result. The material is simply too coarse to separate them, and handing out second, third and fourth place among them would be inventing a difference the documents lack.

The gap between seven and ten points consists almost entirely of one item: the published threshold for the identity check. Three operators state an amount with a clause number, the rest write that they decide for themselves, and that particular difference is the most practically felt one in the whole table. What the three amounts are and where they sit is set out on the page about identity checks and withdrawals.

Section 2 frames verification as a judgement call: documents may be requested, with no amount named. The permit ALSI-202509013-FI1 appears in the public Anjouan register under Zentari Limitada, and the coin list runs to ten entries.

What the bar leaves unmeasured

It measures how much material the operator publishes about itself. It stays clear of judging the games, the payouts or the treatment of customers. An operator that publishes a lot can behave badly, and one that publishes little can behave well.

It also measures nothing about Swedish law. Ten points leaves the gambling unlicensed under spellagen, and a full bar changes nothing in the reach of the exclusion register, which follows from the wording on the page about how far Spelpaus reaches.

And it leaves unmeasured whether the information is true. A published figure is an undertaking rather than an outcome. We have read that it is written down; we have left untested whether it is honoured.

One of the two longest coin lists in the table: thirteen coins, with usdt and usdc among the six given prominence. We read the permit number ALSI-132405048-FI3 here, on the operator’s own page rather than in the Anjouan register; the company behind the brand is Cipher Games Ltd.

A second measure: the supplier list

There is one further item the operator publishes itself and that can be counted. Six of the ten keep their own list of game suppliers, and it can be read entry by entry: Bitcasino.io names 120, Wild.io 80, Empire.io 79, Rainbet 33, Wolf.bet 32 and DuckDice 19.

Four of the six lists were read in archived copies of the operator’s own pages, which means they reflect the range at the copy’s date rather than today. For Rocketpot, Metaspins, Wild Fortune and the tenth row we lack a list we could read, and that counts as a gap rather than as a small range.

The measure is blunt on purpose. A hundred-odd suppliers means more games rather than better games, and the lists contain entries that are hard to count: some names stand for the same studio under two spellings, some are categories rather than suppliers. We have excluded the entries that stayed undocumented in the captured text, and the count is therefore a lower bound.

Three things often ranked that cannot be ranked

Payout time. None of the ten sets of terms gave us a processing time in hours or days. We have timed a payout nowhere either: this site lacks an account with all of the operators and has deposited nothing. Any figure in a payout-time column would therefore be a guess or a quotation from somebody else’s summary.

Bonuses. The offers change more often than a page is updated, and two of the operators moreover write terms that make the amount hard to compare: Rainbet’s terms set a withdrawal cap of 50 dollars on winnings from a promotional code, and Wild Fortune’s cap on winnings from the first three deposits is stated differently on two versions of the same terms page. Both figures fall short of a general withdrawal cap, and we have left them out of the table as such.

Support and service. We have contacted none of the operators. A judgement about response times would rest on a single contact and would say more about the day than about the operator.

The text we worked from is the consolidated wording under SFS 2026:90, in force since 1 May 2026. This is where Chapter 19 on unlawful gambling operations sits, together with 14 kap. 12 § on self-exclusion and the two repealed sections in Chapter 18.

What the six lists have in common

Lay the six lists on top of one another and compare name against name and the total is 182 distinct suppliers. Of those, 85 — almost half — appear on a single list.

Six names appear on all six: BGaming, PG Soft, Pragmatic Play, Spinomenal, Evolution and Hacksaw. Seven more appear on five of six, among them Endorphina, Nolimit City, Play’n GO, Platipus, Slotmill and Spribe. Eleven names appear on four lists, including NetEnt, Blueprint, Playson, Red Tiger, Thunderkick and Big Time Gaming.

The distribution says something about what a count is good for. A core of a couple of dozen studios recurs almost everywhere, and the difference between 19 names and 120 names lies almost entirely in the long tail of suppliers that only one of the six carries. An operator with three times as many names therefore falls well short of offering three times as much of what recurs.

Why the count is a lower bound rather than a figure

The last two names in the core show the problem with counting names. Evolution appears as Evolution on five of the lists and as Evolution Gaming on the sixth. Hacksaw appears as Hacksaw at one operator and Hacksaw Gaming at five.

Counting the rows straight would make those two studios four entries in the tally. We have merged the pairs we could document as the same studio and left standing those that resisted, which makes 182 a ceiling on the number of distinct names and at the same time makes every individual list a lower bound on the number of studios.

To that is added the fact that an entry is at times something other than a supplier. Some rows are categories rather than studios, and four of the six lists were read in archived copies of the operator’s own pages, that is, in the range as it stood at the copy’s date. For Rocketpot, Metaspins, Wild Fortune and the tenth row we lack a list we could read at all, and that counts as a gap rather than as a small range.

A third measure we considered and set aside

In our material every operator carries a set of tags — sportsbook, live-dealer, original-games, vip-program, provably-fair, instant-withdrawal and crypto-only. They suggest themselves as a third measure, since they exist for every row and can be counted.

We leave them unused, and the reason is that they are our own summaries rather than quotations. The tag was set on the basis of how the cashier and the terms looked at the time of reading, rather than taken from a sentence in a document. A measure built on our own tags would measure our reading and then be presented as a property of the operator.

One of the names shows why that is dangerous. Three of the ten carry the tag instant-withdrawal. None of the ten sets of terms contains a processing time in hours or days, and we have timed nothing ourselves — so turning that tag into a column would be putting a number on an item we expressly lack. The same goes for crypto-only, which sits on three rows while a fourth, Wild.io, has a route into fiat through an exchange service the tag misses.

The tags therefore stay as summaries in running text, with the author of the summary stated, and they affect zero bars. How we otherwise separate a quotation from our own label is on the page about how we read the documents.

Why we keep the measures apart from a ranking

The two measures measure different things, and weighing them against one another requires a weight nobody can derive. Is a published verification threshold worth twenty suppliers? The question lacks an answer in every document, and an answer invented here would become the most influential number on the site while resting on nothing.

So the measures stand separately, and the table is sorted by the first without being called a top list. The reader weighs for themselves and can see what each figure rests on. How we handle gaps, archived copies and corrections is on the page about how we read the documents.

What you can measure yourself in a quarter of an hour

Look the permit number up in the issuer’s register rather than trusting the footer. In four of the nine cases the number on the site differs from the number in the register, and the reasons are set out on the page about offshore licences.

Open the verification clause in the terms and read whether it states an amount or a discretion. Then open the withdrawal clause and read three things: the amount, the period and the currency.

Finally read the country list and search for both Sweden and SE. Eight of the 38 lists we read contain Sweden, and the names are on the page about offshore operators. Four answers in a quarter of an hour is four more than a star rating gives you.