The RTP version problem: which build of the slot is your casino actually running?
A modern video slot is not one product. It is a family of certified builds of the same game, each with a different return-to-player figure, and the version you get depends on a licensing decision made between the studio and your casino. This guide explains how that pipeline works, what a 12-point spread costs you in practice, and the method we now use to check a casino's build against the game engine's own configuration — not the label on the page.
- Play'n GO certifies Book of Dead at 96.21%, 94.25%, 91.25%, 87.25% and 84.18%: a 12-point spread that turns a €3.79 long-run cost per €100 wagered into €15.82 on the bottom build (certified RTP levels, verified June 2026).
- Reading the game engine rather than the page label on 17 September 2026, Sweet Bonanza ran at 96.49% at Stake, 7Bit, Gamdom and Rainbet alike, yet Stake's page copy said 96.50% and 7Bit's 96.51%.
- Ladders are the norm, not the exception: Starburst is certified from 99.06% down to 90.05%, Le Bandit from 96.34% to 88.36%, Fire in the Hole 2 from 96.07% to 87.05% (provider flexible-RTP documentation, verified July 2026).
- UK Remote Technical Standard RTS 3C requires the RTP percentage to be made available to the customer in-game (Gambling Commission, page last updated 7 May 2024; read 6 October 2026); lobby pages and marketing copy are not covered.
- A paytable check takes about 30 seconds and four steps: launch the game, open the info screen, find the "RTP" or "return" line, compare it with the provider ladder.
- All engine readings are demo-mode launches; a difference between casinos is strong evidence, identical values across casinos are weaker, and the real-money paytable remains the authority.
Why does one slot have several RTP versions?
Because the studio submits the same game logic with several reel weightings for certification, and each certified build is sold to operators at a different revenue split. The graphics, features and prizes are identical; only the invisible probability weighting differs. The headline figure is the top of a ladder, and a casino can legally license any rung below it.
When a studio finishes a slot, it doesn't submit a single mathematical model for certification. It submits the same game logic with several reel-weighting configurations, each producing a different long-run return, and has each one independently tested and certified. The graphics, features, sounds and even the paytable prizes are identical across all of them — what differs is the invisible probability weighting behind the reels. The result is a menu: a headline build near the top (the number you'll see on review sites, including in our database), and a descending ladder of cheaper builds below it.
The word "cheaper" is deliberate. A slot is licensed to an operator as a revenue-sharing product, and the lower-RTP build keeps a larger share of turnover for the house. From the operator's side of the counter, choosing the 94% build over the 96% build is a straightforward pricing decision — like a shop choosing its margin — except the price change is invisible to the customer standing in front of the shelf.
Who decides which RTP build a casino runs?
The operator, within limits set by two others. The studio defines and certifies the ladder; the casino licenses one rung per market, and large groups may run different rungs on different country-facing sites; the regulator may impose a floor or an in-game disclosure duty. Nobody in that chain has to tell you the number changed. Three parties touch the decision, and only one of them is the studio:
- The studio defines the ladder and certifies each rung with a test lab. It sets the ceiling and the floor, but not which rung you play.
- The operator (your casino) licenses a specific build per market. Some run the top build everywhere as a selling point; others take the discount. Large operators can even run different builds of the same game on different country-facing sites.
- The regulator constrains the choice. Some jurisdictions impose minimum return levels or require the configured figure to be disclosed in-game; others leave the choice free. This is why the same casino brand may pay differently depending on which licence is serving you.
Nothing in that chain requires anyone to tell you the number has changed. The disclosure duty — where one exists — is satisfied by a line of small text inside the game's help screen.
How much does a lower RTP build cost? Book of Dead's 12-point spread
Play'n GO's Book of Dead is the cleanest illustration in the business, because its ladder is unusually deep. The default build returns 96.21%, and certified reduced builds run at 94.25%, 91.25%, 87.25% and 84.18% — a spread of just over twelve RTP points between the version reviewers describe and the version a maximally-tight operator can legally serve.
Twelve points sounds abstract, so put it in money. RTP is the share of total wagers returned over the long run, which makes the house edge simply 100% minus the build:
| Book of Dead build | House edge | Long-run cost per €100 wagered | Relative to the default |
|---|---|---|---|
| 96.21% | 3.79% | €3.79 | the reviewed version |
| 94.25% | 5.75% | €5.75 | ≈1.5× the default's cost |
| 91.25% | 8.75% | €8.75 | ≈2.3× |
| 87.25% | 12.75% | €12.75 | ≈3.4× |
| 84.18% | 15.82% | €15.82 | ≈4.2× |
Every one of those five games is legitimately called "Book of Dead", ships the same 5,000x top prize, and triggers the same free spins. On the bottom rung you are paying more than four times the long-run price for identical entertainment. No session will reveal which rung you're on — variance drowns the signal over any human-scale sample — which is precisely why the build has to be read, not felt.
Book of Dead is deep but not unique. From the ladders in our verified database: Hacksaw's Le Bandit is certified at 96.34 / 94.23 / 92.17 / 88.36%, its stablemate Wanted Dead or a Wild at 96.38 / 94.55 / 92.33 / 88.42%, Nolimit's Fire in the Hole 2 spans 96.07% down to 87.05%, Push Gaming's Razor Shark reaches down to 90.52%, and NetEnt's Starburst ladder stretches from 99.06% to 90.05%. If you want the full mechanics of ladders provider by provider, our configurable RTP guide covers that ground; this page is about verifying the build you've been given.
What do regulators require casinos to disclose about RTP?
In Great Britain, RTS 3C of the Gambling Commission's Remote Technical Standards requires the return-to-player percentage to be made available to the customer, and the in-game information must describe the game as configured; the Malta Gaming Authority imposes a similar duty. Neither requires disclosure outside the game client, and unlicensed operators carry no such duty at all.
The player's one piece of legal leverage is the in-game disclosure rule. The UK Gambling Commission's Remote Technical Standards require that the return-to-player information available to the player from within the game describes the game as configured — meaning a UK-licensed casino running the 94.25% build of Book of Dead must show 94.25% in the game's help screen, not the 96.21% headline. The Malta Gaming Authority imposes a similar duty on its licensees, and MGA-licensed brands make up a large share of the international market. Unregulated and lightly-regulated operators carry no such duty — which is exactly the segment where independent measurement matters most.
Note what the rules do not require: nobody has to display the figure outside the game, nobody has to tell you it differs from the provider default, and lobby pages or marketing copy are typically not covered at all. The regulatory guarantee, where it exists, lives in one place: the game client itself.
How do you check which RTP version you are playing?
Open the game's info screen and read the RTP line. It takes about thirty seconds: launch the slot (demo is fine), tap the "i", "?" or menu icon, search the rules or paytable text for "RTP" or "return", and compare the figure with the provider's certified ladder for that game. A match with a lower rung tells you exactly which build you have been given.
- Launch the game — demo mode is fine where offered.
- Open the info screen: the "i", "?" or hamburger icon, then look for a section named rules, game info or paytable.
- Search the text for "RTP" or "return". The line usually reads like "The theoretical return to player of this game is 94.25%". If ante-bet or feature-buy modes exist, they will list their own figures alongside.
- Compare against the provider ladder for that game (our database lists the certified variants for all 254 slots we cover). If the number matches a lower rung, you now know the build — and can decide with open eyes whether to play it.
How do you read the RTP from the game engine instead of the label?
For Pragmatic Play titles, open the browser's network inspector while the game loads and find the client's initialisation request to the provider's game server; the response carries a plain rtp= field with the value the maths engine was configured to run for that operator's launch. That number is generated by the system that settles spins, which page copy is not.
While building our tracker we found a method that is more authoritative than any printed label — including, it turns out, the casino's own page copy. When a Pragmatic Play game launches, the client makes an initialisation request to the provider's game server, and the response contains the RTP the engine has been configured to run for that specific operator, readable in the browser's network traffic as a plain rtp= field. That value isn't marketing text. It is the number the maths engine itself was handed for the session.
In September 2026 we ran this engine reading for Sweet Bonanza across several crypto casinos, alongside the RTP each casino printed on its own game page. The engines agreed with each other; the page labels did not always agree with their own engines:
| Casino | Page copy says | Engine init reports | Read date | Source |
|---|---|---|---|---|
| Stake | 96.50% | 96.49% | 2026-09-17 | Casino game page vs gameService doInit, demo launch |
| 7Bit | 96.51% | 96.49% | 2026-09-17 | Casino game page vs gameService doInit, demo launch |
| Gamdom | 96.49% | 96.49% | 2026-09-17 | Casino game page vs gameService doInit, demo launch |
| Rainbet | 96.49% | 96.49% | 2026-09-17 | Casino game page vs gameService doInit, demo launch |
To be clear about scale and intent: these are hundredths of a point, not hidden reduced builds, and we don't read them as anything sinister — most likely the page copy was written from a slightly different reference sheet and never reconciled. Stake's page copy overstates its engine figure by 0.01pp; 7Bit's by 0.02pp (7Bit's Sweet Bonanza page also states a "211x maximum win", where the game's actual ceiling is 21,100x — an obvious typo, but one that has sat on a published game page). Gamdom and Rainbet label the engine figure exactly. The lesson isn't that any of these casinos shortchanges players; it's methodological: page labels are copywriting, engine configuration is ground truth, and when the two disagree even by a hundredth, only one of them is generated by the system that settles your spins.
What should you do with this?
Our RTP Version Tracker applies this method on a monthly cycle: for each covered slot and casino we record the figure, how it was obtained (in-game paytable, operator page label, or provider default), and the date it was read. Cells we cannot measure honestly say so — some casinos gate demo play behind registration, and some launch games through aggregators whose traffic can't be inspected. We'd rather show you an empty, explained cell than a copied number.
For your own play, the routine is simpler: know the ladder, read the paytable, and treat any figure that isn't from inside the game — including ours — as provisional. A casino that runs top builds and labels them accurately is telling you something about how it treats players. So is one that doesn't.