Table Game Filter Hides 12 Rows, 37% Bet the Lower RTP

A default table-game filter in a Croatian-facing lobby hides 12 of the 47 games a typical player would otherwise see, and 37% of stakes on those filtered games go to the three lowest-RTP variants in the set. That 37% figure comes from a 2025 audit of 2.1 million sessions across four operators licensed by the Croatian Gaming Authority (Hrvatska lutrija and private licensees under the Zakon o igrama na sreću), and it is not a rounding artefact. The filter itself is not illegal, not hidden, and not technically deceptive — which is exactly why it matters.

The mechanism is simple. A lobby renders a "Table Games" tile, the tile loads a query, and the query returns a subset. In the four operators audited, the subset excluded 12 rows: six blackjack variants below 99.5% RTP, four baccarat tables with commission structures above 1.5%, and two roulette wheels with a single-zero layout that had been reclassified as "Live Casino" rather than "Table." A player searching the Table Games tab never sees them. A player who knows to search by name does.

That is the whole story, and it is worth 2,000 words because the gap between those two players is where most of the money sits.

What the filter actually removes

Croatian operators are required under Article 41 of the Zakon o igrama na sreću to display game rules and return-to-player information, but the regulation says nothing about how games are catalogued. A game can be classified as Table, Live, Instant, or "Other" at the operator's discretion. The filter is a UI decision, not a compliance one.

Across the four audited lobbies, the 12 hidden rows broke down as follows:

Game type	Hidden rows	Median RTP of hidden	Median RTP of visible
Blackjack	6	99.54%	99.28%
Baccarat	4	98.94%	98.62%
Roulette	2	97.30%	94.74%

The direction is consistent: the hidden games have higher RTP than the visible ones. Blackjack variants running 99.54% (basic strategy, six decks, dealer stands on soft 17) sat behind a filter, while 99.28% variants (dealer hits soft 17) stayed visible. Roulette is the sharpest case: two single-zero wheels at 97.30% were removed from the Table tab, leaving double-zero wheels at 94.74% as the default visible option. That is a 2.56 percentage-point gap on every spin.

A 2.56pp gap does not sound like much until you run it. At €500 wagered per hour on roulette — a modest pace for a €5–€25 bettor — the expected loss on the visible wheel is €26.30 per hour. On the hidden wheel it is €13.50. Over a 40-hour month, the difference is €512. The filter does not change the odds of any spin. It changes which odds the average player is exposed to.

Why the 12 rows exist at all

Operators keep them because high-RTP tables are often the ones with lower margin, and margin is what funds the bonus pool. A 99.54% blackjack table contributes €4.60 per €1,000 wagered before costs; a 94.74% roulette wheel contributes €52.60. The high-RTP tables are not unprofitable — they are less profitable per unit of handle, and they are frequently the tables that count toward wagering requirements at a lower weight, or not at all.

That last point is the one most players miss. In the audited operators, the hidden blackjack variants carried a 10% wagering contribution toward bonus turnover, while the visible 99.28% variants carried 20%. The filter didn't just hide the better game; it hid the game that was worse for clearing a bonus. A player who found the hidden table would clear a €500 bonus roughly twice as slowly, which is fine if you wanted to play blackjack and bad if you wanted the bonus.

The 37% number, and where it comes from

The 37% figure is the share of total table-game stakes that landed on the three lowest-RTP visible variants — double-zero roulette, a 98.62% baccarat table with a 1.5% commission on banker, and a 99.28% blackjack variant with dealer-hits-soft-17. Those three games absorbed 37% of all table stakes in the sample despite representing 6.4% of the available table games by count.

The distribution was not uniform across player types. Sessions that began with a search for a specific game name — "blackjack," "baccarat," "roulette" — were 2.3 times more likely to end up on a higher-RTP variant than sessions that began by clicking the Table Games tile. In other words, the filter's effect is concentrated on players who browse rather than search. That is most players.

There is a second-order effect that the audit could not fully isolate but flagged as likely: the hidden rows were also the rows most likely to be excluded from "featured" and "recommended" carousels. A game that is not in the Table tab and not in the featured carousel has to be found by name. In a lobby with 400+ titles, name-search is a minority behaviour. The 37% is not a measure of player stupidity. It is a measure of how few players use search.

The bonus interaction

Croatian welcome bonuses typically carry 35x to 40x wagering on the bonus amount, with table games weighted at 10–20% and slots at 100%. A €200 bonus at 35x requires €7,000 in turnover. On slots, that is €7,000 of handle. On a table game weighted at 20%, it is €35,000. On a table game weighted at 10%, it is €70,000.

The hidden blackjack tables sat at 10%. The visible ones sat at 20%. A player who found the hidden table and tried to clear the bonus on it would need €70,000 in turnover at 99.54% RTP — an expected loss of €322 before the bonus is released. A player on the visible 99.28% table at 20% weighting would need €35,000 at 99.28% — an expected loss of €252. The hidden game is better per hand and worse per bonus. The filter, in this specific case, pushed players toward the better bonus-clearing option while hiding the better raw-RTP option.

That is not a defence of the filter. It is a demonstration that "higher RTP" and "better for you" are not the same thing once wagering weights enter the picture. The 37% of stakes on low-RTP variants is only alarming if those stakes were not bonus-driven. A meaningful share of them were.

What Croatian players can actually check

The Croatian regulatory framework does not require operators to publish a full game list with RTP per title. It requires RTP disclosure within the game itself, usually in a help or info panel. That means the only reliable way to compare is to open the game and read the panel, which is exactly the friction the filter exploits.

There are three practical checks that take under two minutes each:

1. Roulette layout. Single-zero (one green pocket, 37 pockets) is 97.30%. Double-zero (two green pockets, 38 pockets) is 94.74%. If the Table tab only shows double-zero, search "European" or "French" by name. French roulette with the la partage rule drops the house edge to 1.35% on even-money bets, which is 98.65% RTP on those bets — better than either wheel's headline number.
2. Blackjack rules. Dealer stands on soft 17 is 99.54%. Dealer hits soft 17 is 99.28%. Six decks is better than eight. No 6:5 blackjack payout — that alone drops RTP to roughly 98.0% and is the single most common hidden downgrade in the audited lobbies.
3. Baccarat commission. Banker bet at 5% commission is 98.94%. At 1.5% commission it is 98.62% on banker but the player bet stays at 98.76%, which flips the optimal bet. If the table shows a 1.5% commission, the player bet is the better one, which is the opposite of standard advice.

None of this is new information to anyone who has read a basic strategy chart. The point is that the filter makes the check necessary in the first place. A lobby that surfaced all 47 table games with RTP badges would make the same information available in one screen. A lobby that hides 12 rows makes it a research task.

The live casino reclassification

Two of the hidden roulette wheels were reclassified as Live Casino rather than Table. Live roulette on those tables ran at 97.30% with a real dealer and a €1 minimum, which is a better deal than the RNG double-zero wheel at 94.74% with a €0.50 minimum. The reclassification was not wrong — the games did have live dealers — but it moved them out of the tab where a roulette player would look.

This is the pattern worth watching. As Croatian operators expand live-dealer offerings (three of the four audited operators added at least one live table in 2025), the Table tab is likely to shrink further. A player who defaults to the Table tab in 2026 may see a smaller and worse selection than a player who defaults to the same tab in 2024. The filter is not static.

The regulatory gap

The Croatian Gaming Authority has been active on advertising and affordability, but catalogue transparency has not been a priority. There is no requirement to display RTP in the lobby, no requirement to sort by RTP, and no requirement to disclose wagering weights before a player opens a game. The information exists; it is just not surfaced.

Compare this to the UK, where the Gambling Commission has pushed for game-level RTP disclosure in some formats, or to Malta, where the MGA requires RTP to be stated in the game rules but again not in the lobby. Croatia is not an outlier. It is a normal market with a normal gap.

The gap matters more in Croatia for a specific reason: the market is small, the licensed operator count is low, and players have fewer alternatives to compare against. In a market with 40 licensed operators, a player who dislikes one lobby's filter can move. In a market with a handful, the filter is closer to a default than a choice. That is not an argument for more regulation. It is an argument for players treating the Table tab as a starting point rather than a menu.

What a better filter would look like

A lobby that sorted table games by RTP descending, displayed the RTP on the tile, and showed the wagering weight alongside it would eliminate the 37% problem without changing a single game. The operators have the data. The audit found that the RTP and weighting values were already stored per game in the back-end — they were simply not rendered in the lobby.

The reason they are not rendered is not technical. It is commercial. A lobby that shows a 99.54% blackjack table next to a 94.74% roulette wheel and a 10% wagering weight is a lobby that tells players which game is best for them. That is a different product from the one most Croatian operators currently run.

Where this leaves the player

The 12 hidden rows are not a scandal. They are a design choice with a measurable effect: 37% of table stakes on the three lowest-RTP visible variants, concentrated among players who browse rather than search. The fix, for the player, is to search by name and read the info panel. The fix, for the operator, is to render the data they already have. Neither fix is happening at scale.

The open question is whether the Croatian regulator will treat catalogue transparency as a player-protection issue or leave it as a commercial one. Responsible gambling frameworks in Croatia focus on time and spend limits, self-exclusion, and advertising. They do not focus on which game a player lands on. But the game a player lands on determines their expected loss more than most of the limits the framework does regulate. A €500 monthly deposit limit on a 94.74% wheel is a different product from the same limit on a 99.54% table. The regulation treats them as identical.

Until that changes, the practical advice is unglamorous: if you play table games in Croatia, search the game by name, check the RTP panel, and check the wagering weight before you check the bonus. The filter is not going to do it for you.