Live casino vs RNG: what the documents show

A dealt-by-camera table and a machine-dealt round differ in one place this site can actually reach: the record kept about them. Our library holds operator terms, licence certificates and register rows, not game mathematics. So this piece will not give you the house edge of a blackjack shoe or the return of a slot, because no file here contains either figure. What it can do is show what is written down about a live casino across a hundred operator records, where those records stop, and which of the differences sold as a live casino advantage survive a document check.

What a live casino record looks like in our library

The operator library behind this site carries 100 brand records. Each one holds a name, a domain, a licence block, a coin list, a verification summary, a limits block, a restricted-country list and a short tag list. There is no field for game titles, no field for return to player, no field for table rules, no field for dealer studios room by room. That is the whole shape of it.

Forty-seven of the hundred records carry the tag live-dealer. That count is the largest single piece of dealt-table evidence we hold, and it is a tag rather than an audit: it records that the operator’s own pages present dealt tables at the date in the record’s checked field, which for ninety-nine of the hundred records is 26 August 2026. It does not say how many tables, which studio runs them, what the minimum stake is, or whether a table was open on the day you read it. A reader who wants to know whether a live casino lobby is any good has to go somewhere other than this library, and we would rather write that sentence than fill the gap with a guess.

The other tags in the same list are sportsbook (43 records), vip-program (37), provably-fair (21), original-games (20), crypto-only (14) and instant-withdrawal (12). Nothing in that set describes a game’s behaviour. A page that claims to compare a live casino floor with an RNG lobby on the strength of these tags is comparing labels.

Studio strings, and why a count of them is not a count of companies

Thirty-three of the hundred records carry a providers list copied from the operator’s own catalogue page. Those lists are where the word live actually appears as part of a supplier name. Bitcasino.io lists Evolution, Ezugi, Bombay Live, Live Solutions, Live88, LiveSpins, Microgaming Live and Winfinity in the same array as slot studios. Livecasino.io lists Evolution, Ezugi, Bombay Live, Live Solutions, Live88, Livespins and Pragmatic Play Live. Roobet’s list carries OneTouch and OneTouch Live as two separate strings. TrustDice lists Absolute Live alongside Evolution and Ezugi.

Two strings, one company, is the normal case rather than the exception. The same catalogues give us Evolution and Evolution Gaming, Pragmatic Play and Pragmatic Play Live, Microgaming and MicroGaming, Nolimit City and NoLimit City and No Limit City, TrueLab and True Labs and Truelab and TRUE LABS, Voltent and VoltEnt and VoltENT, Play’n GO and Play’n Go and one entry with a curly apostrophe. Counting rows in those arrays produces a number; it does not produce a number of companies. Any sentence of the form this live casino runs N studios built from our data is a string count wearing a company count’s clothes.

The scoring file makes the same point in its own way. Its weights list includes providers at 1 point, and its _unscored array names providers as a component that is never awarded. Catalogue length earns nothing in our rating, and the file says so in two lines rather than in prose.

One brand is a live casino by name, and its licence has a twin

The clearest dealt-table fact in the library is not about tables at all. Livecasino.io appears as its own record, tagged live-dealer, with the Curaçao Gaming Authority licence number OGL/2023/111/0069, held by Moon Technologies B.V., status Active, granted 1 July 2024, read in the certificate portal on 26 August 2026. Bitcasino.io carries exactly the same licence number, the same holder, the same company number 152185 and the same grant date.

So the brand that puts the words in its domain and the brand that does not are one licence row. That is a checkable statement with a date on it, and it says more about how the market is organised than any claim about dealer quality. The same shape repeats elsewhere: OGL/2024/1307/0748 and company number 162990 cover 7Bit Casino, KatsuBet and Mirax, of which KatsuBet and Mirax are tagged live-dealer. We set out the full set of these groupings in the piece on which holders sit behind the brand names.

The verification wording for those three is not identical, which is worth noting because it is the part that touches a player. Mirax’s terms describe KYC as an obligatory procedure covering proof of identity, address and payment, clause 7, read from the operator’s own terms page. KatsuBet’s FAQ states that playing does not require account verification but that KYC may be requested at the operator’s sole discretion before any payment. One licence, one company number, two different sentences about documents. That divergence is the most useful thing a live casino comparison can surface, because the clause is what decides whether a win leaves the account.

What a live casino and an RNG game share: the operator’s clauses

Everything our files can say about the moment a game pays out applies to both kinds of game, because the clauses are written about accounts and not about tables. Bitcasino.io’s terms allow additional checks on withdrawals above 2,500 euro or the crypto equivalent, clause 6.6. Livecasino.io’s terms carry the same 2,500 euro figure at the same clause number, and add that random account health checks may happen at any time. Empire.io sets its additional-check line at 2,000 USDT, clause 5.4. Rocketpot names 0.05 BTC or 2,500 US dollars, clause 11.4, and gives 30 calendar days to supply documents.

Ceilings work the same way. Bitcasino.io states a weekly withdrawal limit of 1,000,000 USDT at clause 6.10. Mirax states 10,000 euro a week at clause 8. Oshi Casino states 4,000 euro a day at clause 12. Wild Fortune states 15,000 euro a month at clause 11.8, and its terms add that Bitcoin players must submit valid identity documents before playing, with deposits and bets blocked until that is done. None of those numbers changes when you move from a slot to a dealt table, and none of them appears in a game’s paytable.

This is the practical asymmetry. The game layer is where the marketing sits and where our evidence is thinnest. The account layer is where our evidence is thickest and where a session actually ends. If you want the account layer laid out on its own, the piece on where anonymous play stops according to the terms does that, clause by clause and brand by brand.

The arithmetic we are not doing

A fair question at this point is why we do not simply compute the comparison. Suppose a dealt table returns a share R1 of stakes over a long run and a machine-dealt game returns R2, and suppose a player stakes an amount S in total. The expected difference in returns is S multiplied by the difference between R1 and R2. The structure of that sentence is not in dispute. The inputs are: our files hold no value for R1, no value for R2, and no per-round history from which either could be estimated.

If we substituted figures to make the sum look finished, those figures would be invented, and the page would then be a guess with a multiplication sign in it. Where we do put a number in a worked example anywhere on this site, it is labelled as a substitution and not as a reading. The honest output here is the shape of the calculation plus a statement of which cells are empty, and the empty cells are ours, not the operators’.

The same limit applies to session length, hand speed, dealer count and table minimums. Those are all measurable quantities. None of them is measured in any file we hold, so none of them is printed as a figure on this site.

It is worth being blunt about what that rules out. We cannot tell you that a live casino table is tighter or looser than a reel game, we cannot rank one dealer supplier against another, and we cannot support any claim that one kind of game pays back more than the other over a session of a given length. Every one of those sentences needs a number we do not hold. What we can do is put the licence row, the verification clause and the withdrawal ceiling for each brand next to each other, which is what the rest of this site does.

Where the dealt-table label does carry weight

Three things follow from the tag that are worth keeping. First, a dealt table is run by a third party for the operator, and the supplier strings in our catalogue captures show how concentrated that layer is: Evolution in some spelling appears in the provider lists of BetFury, Bitcasino.io, Bitsler, Chips.gg, CryptoLeo, Flush, Gamdom, LTC Casino, Lucky Block, Rainbet, Roobet, Sportsbet.io, Thrill, TrustDice, Wild.io, Fairspin, FortuneJack, Windice, DuckDice, Wolf.bet, SolCasino, Livecasino.io, Empire.io, Instant Casino, Betcoin.ag and Playbet.io. That is a supplier-side fact, captured from operator pages, and it is the kind of fact a catalogue can support.

Second, the provably-fair tag and the live-dealer tag are different claims and sit on different records. Twenty-one records carry the fairness tag; forty-seven carry the dealt-table tag; the sets overlap but neither contains the other. A dealt table is verified by being watched, a seeded game by being recomputed, and our piece on how a seeded round is recomputed covers the second of those. Nothing in our files lets us recompute a dealt shoe, so a live casino round is, in the strict sense, the less checkable of the two.

Third, the licence row is indifferent to the distinction. Whichever kind of game is running, the certificate names a company, a company number, a licence number and a status, and that is what the piece on reading a licence certificate line by line is for. Start from the tag-by-tag list of top no KYC casinos if you would rather see the records first and the reasoning second.

What would settle the comparison

A comparison of a dealt table against a machine-dealt game would need three inputs we do not have: the published mathematics for each title, a per-round log long enough to test it, and a dated capture of the lobby proving the title was available where it was claimed. Two of those three would have to come from the supplier rather than the operator, and none of the three is in the three files this page is allowed to read.

Until they are, the useful reading of a live casino badge is narrow and worth stating plainly: it tells you the operator presented dealt tables on the day we looked, it tells you nothing about the odds at those tables, and it changes none of the verification clauses, ceilings or licence rows that decide what happens when you try to take money out. The rest of this section works on the layers where the evidence is actually written down, and our record of payout clauses read from operator terms is the closest thing here to a measured result.