Anybody can list fifty casinos in an afternoon. The reason our lists are short is that four things have to be true before an operator appears on one, and we would rather publish a short honest list than a long decorative one.
Each one is harder to pass than the last, which is why the fourth is the one most sites skip.
This is the part worth reading, because it is where most sites quietly round up.
Every figure on this site carries the place it came from: a live lobby search with a date, an operator’s published terms, a studio’s own catalogue, or a keyword tool with the month of the reading. Where a number is a model rather than a measurement — the step tables are — the assumption is printed next to it.
A return-to-player figure for Chicken Road, because nobody publishes one. A maximum multiplier, for the same reason. A licence we could not locate in a register, stated as though we had. The gaps are deliberate and they are the most reliable signal on the site.
Today there are none. When outbound operator links arrive, arrangement pays for the work; it does not decide what appears here. A casino gets a recommendation after clearing checks, and a casino that fails one gets that written on its page whether or not there is a commercial relationship behind it. The clearest evidence we can offer is the Mostbet licence note, which is on our own page for an operator we link to.
A method is only worth reading if it catches something. These are the errors our own checks found in our own pages, in the order they were found.
It publishes thirty-four of them. We had crawled the shop-window domain, where game pages sit behind a redirect, and concluded from an empty result that the data did not exist. The catalogue was on a different domain and opened freely. The claim was on nine pages.
One does. We read all forty pages and the line «RTP: Configurable according to operator settings» appears on the Chicken Road page alone.
It is 90%, on the studio’s oldest title. 94% is the floor of the chicken line, not of the catalogue — a distinction we had collapsed.
The worst of the six. We called Ice Fish a run across a frozen lake with a cash-out before the ice cracks. It is a betting wheel with no steps in it. The capture proving that was in our own files, unopened, while we wrote the page from the cover art.
It does not. Wheel Out climbs continuously and carries a pickup that halves the accumulated multiplier. A rule built for a curve that only rises does not apply to one that can fall.
A consequence of the first error. Once the INOUT catalogue was read properly, publishing a return rate turned out to be the norm in this genre rather than the exception.
Five of the six are the same mistake: treating an absence of evidence as evidence of absence. We did not find a figure, so we wrote that no figure existed. We did not have a capture, so we described the game from its name. Both are the cheap version of research and both read exactly like the expensive version on the page.
A negative claim now requires the search that would have found the positive one, named on the page. «We looked in this specific place on this date and it was not there» is publishable. «It does not exist» is not.
Because a site that has never corrected anything has either never checked or is not telling you. Publishing the list costs us something, and that is precisely what makes it worth reading.
Silently edit. A corrected claim keeps a line on the page saying what it used to say and why it changed, because the readers most affected are the ones who read the wrong version.
Before believing that a page is missing, we request a slug that cannot exist. If the invented address returns a proper 404, then a 404 on a real address means the page is genuinely gone. If the site answers everything with a 200, every result it gave us is worthless. It takes one extra request and it is the difference between a finding and a guess.
Repeatable in about five minutes, and it is the check that most often returns something the operator does not mention.
The operator’s own site is the claim we are testing. A number repeated by another affiliate has already passed through somebody else’s reading.
Not a search box on a third-party site. The file itself, dated, saved, so the check can be repeated later against the same evidence.
They fail differently. A number can be absent while the group appears under a different entity, which is exactly what we found in one case.
Neither means much alone. An expired date with an «assessment in progress» status is a backlog; the same status on a current licence is not.
This is the step most reviews skip and it changes the meaning entirely. When four in ten entries carry the same flag, the flag describes the regulator rather than the operator.
Three of four checks found something the operator does not advertise: two licences past their expiry date, one held by a different company than the brand implies, and one published number that appears nowhere in the current register at all. None of those is a scandal. All of them are things you would want to know before choosing where your balance sits.
Whether the money comes back. A licence is a statement about permission, not about behaviour. The only evidence that matters on that question is a completed withdrawal, and it is the evidence this category is worst at producing.
Because «licence expired» without «and so has four in ten of this register» is technically true and functionally misleading. A fact that reliably produces a wrong conclusion is not a fact we have published responsibly.
Fourteen research passes across review platforms and complaint services, half of them adversarial. Not one collection survived checking, and the reason says more about this category than any single review would.
Player reports of money actually arriving or not: the method, the amount, the timeframe, whether verification was demanded, whether a payout was refused. The one thing every operator page on this site openly lacks.
Trustpilot and Casino Guru return an error to automated requests. Everything then arrives through a search engine’s summary of the page rather than the page — and a summary of a review, presented in quotation marks, is not a quote. Thirteen of fifteen findings on one operator rested on exactly that.
A quoted line that did not appear on the page it cited. An entire set of complaints attributed to one brand that belonged to a different company. Six reviews taken from a site whose own disclaimer says it is not a casino. Cases reported as unresolved where the source showed them settled, with the operator’s reply visible.
One operator appeared as 1.4, 1.5 and 2.7 out of five in three different passes, while the live page showed 2.4. Two of those three figures traced back to affiliate blacklist pages rather than the review platform they were credited to.
Because the alternative is to print plausible-looking testimony we cannot stand behind — which is the exact practice this site was built to point at. A blank here is not a gap in effort. It is the honest state of what open sources can support.
Complaints against one operator on a dispute service ran 21 resolved, 2 open, 1 rejected across the visible page. Another operator shows 2.4 out of 5 from 84 reviews, of which 69% are one star. A third shows 8.4 out of 10 from just 14 reviews while carrying 46 complaints on the same platform — a gap wide enough that the score should not be read as a verdict. And one documented case took about a month of verification before a payout of $144.79 completed.
Our own deposit and withdrawal, documented. Until then everything this site says about payouts describes published terms rather than observed behaviour, and we would rather repeat that than let a table of corridors imply we watched money move.
A method page that only lists strengths is a sales page with numbered headings.
A licence is permission, not conduct. An operator can hold a current licence, publish a corridor, and still take three weeks over a withdrawal because a verification queue is understaffed. Nothing we check would see that coming.
Every finding here carries the date it was made. A register is republished, a lobby changes supplier, a corridor is quietly revised. A check from three months ago is a historical record, which is why the dates are on the page rather than in a footer.
Where an operator publishes no contribution table, no rupee corridor and no verification requirement, we can report the absence and nothing more. An absence is real information, and it is not the same as knowing what the rule is.
Whether the money comes back. We have not completed a withdrawal from any operator on this site, so every payout statement here describes published terms rather than observed behaviour. That gap is the largest one in our method and we would rather name it than let four green ticks imply otherwise.
Reading forty catalogue pages, searching two registers and capturing three games took longer than writing a fifty-casino list would have. That is the trade, and it is why this site covers a handful of operators rather than a directory of them.
A documented completed withdrawal in the local currency, or a documented failed one. Both are worth more than every other check on this page combined, and both are the evidence this category is worst at producing. If you have either, we want it.
Four checks, in order, with the failures published rather than deleted. Every number carries its source and every model states its assumption. That is the whole method, and it is deliberately boring — the interesting part is what it excludes.
