Dice, limbo, plinko, mines: house edges compared

Dice, limbo, plinko, mines and keno look alike from casino to casino, but each casino runs its own version and sets its own house edge, the share of every bet it expects to keep. This post compares what Housebets, Stake and Gamdom publish about these games with the return to player (RTP) the tracker observed in Housebets’ and Gamdom’s public bet feeds over the 30 days to 8 Nov 2026, 14:22 UTC. It also shows why one name can cover different rules, pay tables and edges.

House edge and RTP in brief

RTP is the share of stakes a game pays back as prizes. The UK Gambling Commission, writing about gaming machines, calls it “an average achieved over a significant number of game plays and not each time” the machine is played. The house edge is the other side of the same number: the percentage a casino “would expect to keep, on average, from each hand or spin”. A 99% RTP is a 1% edge. As an example, $10,000 wagered at a 1% edge is expected to cost $100 in the long run, though any one session can end far from that.

Originals are a casino’s in-house games. Most let the player set the risk: the win chance in dice, the target in limbo, the number of mines. A multiplier is the payout divided by the stake. Volatility describes how unevenly a game pays; high-volatility settings pay rarely, but large.

No setting removes the edge. GambleAware NSW explains that the house edge “means the casino takes a small percentage of every bet. While you might enjoy occasional wins, over time you are likely to lose. Big wins are rare and the longer you play, the more you lose.” Gambling is for adults (18+), and the responsible gambling page covers limits, warning signs and free help.

What the casinos publish

Housebets’ fairness page, with an editorial update dated 11 October 2026, gives the formula behind its in-house dice, limbo, mines, plinko and keno, among other games. Its dice multiplier is 99 divided by the win chance in per cent, “so every bet returns 99%”. In limbo, “every target returns 99%”; in mines, “every board and every cash-out returns 99%”. Every plinko table “returns exactly 99%”, and every keno paytable returns 99% “at every risk level and every number of picks”.

Stake’s provably fair page writes the edge into its formulas: limbo uses “houseEdge of 0.99 (1%)” and its Drill game “houseEdge of 0.98 (2%)”, so its originals don’t all carry one edge.

Gamdom’s help centre gives a rounder figure for its originals, including dice, plinko, mines and keno: RTP “around 99%”. It adds that part of “the remaining percentage” goes to rewards, and that a “Zero Edge” pays a small portion of bets back on certain games “for the first $5,000 wagered”. Game by game, the tracker lists Gamdom’s dice as advertised at 99.88% RTP, a 0.12% house edge, and the other games in the table at 99.00%, a 1.00% edge.

One name, different games

Rules differ too. Housebets’ limbo runs “between 1.00× and 1,000,000×”. Stake’s mines takes 1 to 24 mines, Gamdom’s is played on a 5x5 grid, and Housebets offers boards of 25, 36, 49 or 64 tiles. Stake’s plinko has 8 to 16 rows of pins. Stake and Housebets both draw ten keno numbers from 40; Gamdom offers four keno risk modes, “Classic, Low, Medium, or High”, each with its own payout table. Even at the same edge, a different table spreads returns differently, between frequent small wins and rare large ones.

Roulette shows how far one name can stretch. Stake’s is “derived from the European version” of the game, with 37 pockets, and Housebets prices its single-zero wheel at 97.30%. On CS2 case sites the game prices differently: the tracker lists CSGOEmpire’s and CSGORoll’s roulette at an advertised 93.34% (a 6.66% edge) and 93.30% (6.70%).

Observed RTP at Housebets and Gamdom

These are the most-played dice, limbo, plinko, mines and keno rows on the originals RTP page over the 30 days to 8 Nov 2026, 14:22 UTC. Observed RTP is what the bets the tracker saw paid back, divided by what they staked; the 95% interval shows how far chance alone could move it. Tracking began in November 2026, so the window holds only the days since then.

In every row but keno at Gamdom, the advertised RTP sits inside the interval: on these samples, those games can’t be told apart from ones paying exactly as advertised.

Why limbo and mines swing more

Part of the difference is the sample itself. NIST’s statistics handbook gives two rules: “As N increases, the interval gets narrower”, and noisy data “are going to generate wider intervals”. For dice and limbo, Housebets has more bets and the narrower interval. Keno runs the other way: Gamdom has more bets but the wider interval, so its keno results are the noisier.

Limbo and mines are noisy because they can pay huge multipliers. In Housebets’ limbo, a target t is reached “with probability 0.99 ÷ t”, so the higher the target, the rarer the win. In mines, each safe tile raises the multiplier, and Gamdom notes that more mines mean “bigger potential rewards”. Both games sit near the top of the tracker’s highest multipliers over the same 30 days:

A huge multiplier can turn a small stake into a large payout: on 6 Nov 2026 at 21:34 UTC, a $1.50 limbo bet at Thrill paid $150,000 at 100,000×. Payouts like that are rare by design. One can lift a sample’s observed RTP above the advertised figure; without one, the reading can sit below it. The post on slot volatility explains why big multipliers are rare in slots too.

The keno flag at Gamdom

One row in the table is flagged. Keno at Gamdom had an observed RTP of 88.29% over 1,359,664 bets, with a 95% interval of 77.79% to 98.80%, so its advertised 99.00% sits just above the range.

A flag describes a sample, not a game. As NIST’s handbook explains, if many samples are collected, “in the long run about 95 % of these intervals would contain the true mean”; the rest miss by chance. The originals page holds many rows, so a few misses are expected even if every game pays exactly as designed. The feed is a sample too: it shows some bets, not all, and an interval can’t correct for what a feed leaves out. The flag is a reason to keep watching the row, not a finding about the game. The post on what a sample can tell you explains intervals further.

What the numbers can’t show

The figures cover only the bets public feeds showed and what those bets paid back. They can’t show what any one player ends up with. That also depends on rewards: Gamdom says part of “the remaining percentage” goes to rewards, and its “Zero Edge” pays a small portion of bets back.

Every round is independent: “Past losses don’t increase your chances of winning,” as GambleAware NSW puts it. At 1% or at 0.12%, the house edge means players as a group lose over time.

What to watch next: whether keno at Gamdom stays outside its interval as the 30-day window fills, and whether the wide limbo interval at Gamdom narrows as its sample grows.