The problem we set out to solve
Casino comparison sites rank by commission. That is not a cynical reading, it is the business model: an operator pays more, it appears higher, and the ranking is presented as editorial judgement. Readers have adapted by distrusting all of it, which is a reasonable response and leaves them with no way to compare anything.
RakeRadar takes a narrower approach. We rank on one measurable thing, the cost of placing a bet, and we generate the tables from a database rather than assembling them by hand. An editor cannot move a casino up a list here, because no editor builds the list.
That constraint is the product. It limits what we can cover, and everything we do cover is checkable.
What we measure
Reported return to player. The figures games declare, drawn from catalogue data rather than operator marketing. Across our sample the median is 96.36% and 92% of games sit inside a single percentage point, which tells you most RTP comparison is noise.
On-chain reserves and flows. Crypto casinos settle on public blockchains, so their balances and payouts are countable from outside. This is the only category of casino data that does not require trusting the operator.
Spread between builds of the same game. The same title can run at different returns at different casinos. We found gaps above two percentage points, which is larger than any difference between operators in the overall ranking.
What we do not measure: conduct, dispute handling, support quality, or whether a withdrawal you personally request will arrive. Those matter enormously and no method available to us reaches them.
The data we refuse to use
We track millions of public bet records and deliberately do not use them to calculate returns.
Those feeds are weighted toward wins, because casinos surface notable results and collectors capture what is surfaced. Running the arithmetic across our own feed data produces an implied return above 100%, which would mean the house pays out more than it takes in. That is not a rounding problem, it is a biased sample.
Sites quoting a “real RTP” derived from live bet feeds are quoting a number that cannot be correct. We would rather publish fewer figures than publish that one, and we say so on every page where the question could arise.
Who writes this
RakeRadar is a small independent operation. We are not a media group, we do not employ a content team writing to volume targets, and we do not accept guest posts or sponsored articles.
Every number in a table comes from a query. Every editorial judgement sits in prose next to the number it interprets, so you can accept the data and reject our reading of it. Where we are uncertain, the page says so rather than rounding uncertainty into confidence.
If you find an error, tell us. Corrections are cheaper than reputations. Our contact page has the details, and our methodology sets out exactly how each table is produced.