Dice Game Guide: Crypto Roll-Under Odds, Classic Dice Probability and What 20,000 Sessions Show

A dice game in a crypto casino is usually "roll-under dice": you choose a target, and you win if a random number from 0 to 100 comes in under it. Your win chance equals the target, and the payout shrinks as the chance grows. With the 1% house edge we modelled, a 49.5% chance pays 2x and a 10% chance pays 9.9x, and every choice loses about 1 USDT per 100 USDT bet over time.

This guide covers how crypto dice works, the full win-chance and multiplier table, classic two-dice probability, and what happened when we ran 20,000 bankroll sessions at three different targets.

How the crypto dice game works

Crypto dice replaces physical dice with a random number, usually shown with two decimals. Playing dice this way takes three choices:

1. Stake, for example 1 USDT.
2. Target and direction. "Roll under 25" wins if the result is below 25. Many games also offer "roll over".
3. Roll. If you win, you're paid stake × multiplier. If not, you lose the stake.

Because you set the target, you set the probability. That makes dice the most transparent game in a crypto casino: the win chance is printed on every bet.

In provably fair dice, the roll comes from a hash of a server seed (committed before you bet), your client seed and a counter called a nonce. One major operator's implementation notes explain how the hash output is turned into a number, so you can recompute each roll afterwards.

The odds and the math

The multiplier is set so that the house keeps a fixed share of every bet:

The expected return per bet is then the same at every target:

The pattern is simple: halve the win chance and you roughly double the multiplier. Want a different target? Our dice odds calculator gives the win chance, multiplier and expected loss for any number you choose.

What our simulation shows

1 million rolls per target: the RTP holds

We rolled 1,000,000 simulated dice results at five targets and measured how much was paid back (RTP, return to player).

Every observed RTP landed within about 0.06 percentage points of 99%. Over a million rolls, the edge shows up exactly as the formula predicts, which is the law of large numbers that introductory statistics describes as probability being a long-run relative frequency.

20,000 bankroll sessions: same edge, very different rides

Next we gave 20,000 simulated players a 100 USDT bankroll each and had them bet 1 USDT per roll at one fixed target until they either doubled to 200 USDT, went bust, or reached 10,000 bets.

Reached 200 USDT Went bust

What the numbers show:

• The high-risk target was the most likely to double. At under 10 (9.9x), 43.94% of players reached 200 USDT. Big wins move a bankroll fast in both directions, so sessions ended quickly: median 882 bets.
• The "safe" target never doubled. At under 90 (1.1x), 0% of players reached 200 USDT within 10,000 bets. Wins of 0.10 USDT can't outrun a steady 1% drain; 57.28% busted and the rest were still grinding.
• Even-money was in between. At under 49.5 (2x), 8.65% doubled and 66.28% busted.

The reason is time. Each 1 USDT bet costs 0.01 USDT on average, so 10,000 bets cost about 100 USDT, the whole bankroll. Low-variance targets give the edge more bets to work with before anything dramatic happens.

Classic dice probability

Not every dice game is crypto roll-under. With physical six-sided dice, each face has a 1 in 6 chance (16.67%). With two dice there are 6 × 6 = 36 equally likely combinations, and totals in the middle can be made more ways. We checked the exact figures with 1,000,000 simulated rolls:

Probability here is favourable outcomes divided by total outcomes, as Britannica's guide to gambling odds puts it. These totals drive the casino dice game craps; see our craps odds guide for how they turn into a 1.41% house edge on the pass line.

Dice game strategy: what works and what doesn't

What doesn't work:

• Martingale. Doubling after each loss at 2x wins small amounts often and loses everything occasionally. The Wizard of Odds explains why it doesn't dent the house edge.
• Chasing "due" numbers. Each roll is independent. A run of high results doesn't make a low one more likely; that's the gambler's fallacy.
• Switching between under and over. Both sides carry the same edge.

What does help:

• Choose your volatility on purpose. Our sessions show low win chances give a better shot at a fast double and a faster bust; high win chances give long, slow sessions.
• Bet small relative to your bankroll. At 1% per bet you get many rolls for the same expected cost.
• Set a stop-loss and a win goal before you start. Our survival table shows sessions end in very different ways depending on target.

Playing dice games online with USDT

Dice is native to crypto casinos, and USDT keeps your bankroll steady in dollar terms while you play. A bankroll in BTC or ETH adds market swings on top of game swings.

• Fees matter for small bankrolls. A network fee can cost more than the house edge on dozens of bets. Our USDT payments guides explain networks.
• Verify your rolls. Set your own client seed, note the server seed hash, and check a few results after rotating the seed.
• Try free dice games first. Free online dice games and demo modes let you learn targets and multipliers without risk, as long as you remember a demo can't predict anything.

Like dice's transparent math? Crash games use the same 99 ÷ x formula for the chance of reaching a multiplier, and our Mines odds guide shows another provably fair game where you can compute every probability. All dice guides live on our dice games hub.

Common mistakes

• Believing a 90% win chance is "safe": our simulation shows it never doubled a bankroll and busted 57% of the time.
• Raising stakes after losses to win back faster.
• Ignoring the cost of volume: 10,000 bets of 1 USDT cost about 100 USDT on average.
• Treating a free demo session as evidence for a system.