Two questions we ask about every game
The first question is: is the game in the Dukat.bet lobby on the snapshot day? If we cannot find it in the relevant section that day, it does not enter the list – no matter what is written about it elsewhere. The current snapshots are dated 12 October 2026.
The second question concerns the key. Has someone credible published the figure we sort the list by? For new releases it is the release date; for other lists it is RTP, max win or min. bet. In the three lobby lists the key is the position in the operator tab itself, so the lobby is the source.
A game that passes the first question but not the second does not vanish from the table without a trace. We place it at the end with the note “not published”, or mention it in the part about games outside the snapshot – as fits the list.
Two questions. No third one about our own taste.
Where the numbers come from, and what happens when the maker is silent
The studio carries the most weight. The maker’s game page, its release announcement and the in-game rules or paytable all come from whoever built the game, so the tables label them “studio (official)”.
If the studio does not state a figure, we turn to independent sources. One is not enough; we need two that do not copy each other and agree on the same value. Only then does the number get the label “2 independent sources”. When two sources disagree, we do not pick the more convenient one.
If we find neither, we write “not published”. No estimate, rough rounding or value borrowed from a similar game goes in its place.
RTP adds one more layer: a single game can come in several payback versions from the studio. The RTP versions page explains how we handle this and how to find the version in the game. Dates work similarly – global launch, early access and regional versions are covered under release dates.
The stages of a snapshot in one table
| Stage | What exactly we do | What it does not imply |
|---|---|---|
| Lobby record | We note which games are in the section or tab on the snapshot date | That the game will stay in the lobby |
| Choice of key | We set the single figure the list is sorted by | That other features of the game do not matter |
| Source check | We look for the figure at the studio, then in two independent sources | That the casino runs exactly this version of the game |
| Sorting | We sort by the key; tie = alphabet, blank = end | That a higher position is a better choice for you |
| Row notes | Each game gets a short factual text and its source type | That it is a review or a recommendation |
| Date | We label the snapshot with the day it applies to | That the figures have not changed since |
| Correction | We re-check a reported figure and issue a new snapshot if it is wrong | That errors never happen |
The right-hand column deserves as much attention as the middle one: every stage guards against something, yet none predicts a game’s outcome.
Equal values, missing values
Ties happen more often than you might think. Two new releases come out on the same day, two games share the same max win or the same min. bet. Then the alphabet decides, by game name – a rule that favours no one and takes seconds to check.
Rows without a key value go to the end, again alphabetically. It is not a penalty; there is simply nothing to compare them with.
Sample calculation with fictitious numbers – not Dukat.bet data: imagine a list sorted by max win with four games. “Comet” states 10,000×, “Atlas” and “Birch” both state 5,000×, and “Rainbow” has not published its cap. The order will be: 1. Comet, 2. Atlas, 3. Birch, 4. Rainbow. Atlas comes ahead of Birch only because “A” is earlier in the alphabet; Rainbow is last because we do not know its number, not because it is worse.
When the key is the lobby order itself
The “Top Slots”, “Top Live Casino” and “Top Instant” lists work differently. Here we do not set the order – we take it from the tab the operator put together, exactly as it stood on the snapshot date. We move or drop nothing on a whim.
Such an order is the operator’s curated selection, not a player count or play statistics. It can favour one studio – in “Top Slots”, for example, the first eight places belong to Playson’s Hold and Win games – and we simply record that. The popularity signals page discusses what “popular” could mean and why we stick to the dated tab.
We do not use lobby labels such as “New Games”, “High RTP” or “Jackpot” as data. On inspection they turned out not to match what they promise.
One caveat: the Dukat.bet lobby varies with the country you connect from. Our snapshots match a Slovak connection; from Hungary we saw, for example, a different Top Live Casino tab. That may explain a mismatch.
A new date instead of a silent edit
A snapshot is valid for the day it carries. When the order in a tab changes, a new game arrives or a studio revises a figure, we do not overwrite the table under the old date; we publish a new snapshot with a new date.
That is a principle, not a calendar. We do not promise a fixed refresh rate; we only promise that the date in the header will always match the table’s contents.
If a game leaves the lobby meanwhile, the next snapshot lists it among games outside the snapshot, so its absence is explained.
Five things that never affect the order
Stars and points. A grade looks precise, but behind it sits someone’s impression that cannot be checked.
Voting. Polls measure who turned up to vote, not a property of the game.
Paid positions. A place in a list is not for sale – neither to a studio nor to a casino.
“Feel of the game”. Graphics and sound are a good reason to pick something yourself, but they have no place in a ranking.
Player statistics. We do not have them, so we do not claim which game is played the most.
A position in a list is therefore not an invitation to play. It is the result of one sort on one day.
How to report a figure that looks wrong
Have you noticed that a game has left a tab, a studio has changed its RTP or a release date does not match the maker’s announcement? Drop us a line via contact – the game name, the list and a source link if you have one will do.
We check the report against the same ladder of sources we use to build a snapshot. If the error is confirmed, we issue a corrected snapshot with a new date. Otherwise the figure remains, though we review the source once more.