Crypto Dice in 2026: The Formula Behind Every Roll

Crypto dice is the oldest provably fair game there is, and still one of the cleanest. You pick a number, choose to roll over or under it, set a stake, and the game rolls a number between 0 and 100. That's it.

What makes it interesting is that you set your own odds. Want a near-certain small win? Roll under 95. Want a long shot? Roll under 2. The payout adjusts automatically, and it does so with a formula you can check in ten seconds.

We popped the hood on that formula, the seed system that proves each roll, and the betting systems people bolt on top. One part of the game you control completely. Another you control not at all. Knowing which is which saves money.

Where to play crypto dice in 2026

Dice is almost always an in-house "original" at crypto casinos. Originals tend to run at low, published house edges and accept tiny stakes in any coin the cashier supports. The list below draws on our full Haxies Index, with partner casinos featured first and badged. The Index scores bonus clearing cost, stated payout speed and licensing cover across 17 casinos. It does not score dice games specifically.

In our brand data, two casinos explicitly confirm provably fair originals. BC.Game lists 74 of them, including a Crash game at a stated 99% RTP. Betpanda also markets provably fair originals. For the others, the data we hold doesn't list in-house games, so search the lobby for "Dice" before depositing.

TrustDice has "dice" in its name, but our brand sheet doesn't list its game menu beyond 1,500+ games from 60 studios, so we won't claim specifics. The TrustDice review covers what the operator does publish, including stated instant payouts and a Curaçao licence number.

The highest Index scores on our list of the best crypto casinos go where offers are cheapest to clear and stated payouts are fastest. Those are good traits for the place you keep a dice balance. The dice edge itself, though, you read from the game.

How does crypto dice work?

Crypto dice generates a random number from 0.00 to 99.99. You choose a target and a direction: roll under or roll over. Your win chance is the share of numbers on your side of the target. The payout multiplier is (100 minus the house edge) divided by your win chance. A 1% edge at 50% chance pays 1.98x.

That's the snippet answer. Now the detail.

Say you pick "roll under 50". Winning numbers are 0.00 to 49.99, which is 5,000 of the 10,000 possible results, so a 50% chance. A perfectly fair game would pay 2x. With a 1% edge, it pays 99 divided by 50, which is 1.98x.

Pick "roll under 10" and your chance is 10%. The payout is 99 divided by 10, which is 9.9x. Pick "roll over 2" and your chance is 98%, so the payout is 99 divided by 98, about 1.0102x.

Roll over versus roll under

They are the same game seen from opposite ends. "Roll under 25" and "roll over 75" both win on 25% of rolls and pay the same multiplier. Some games show the range as a slider, others as a number box. The maths underneath is identical.

Watch the boundary, though. Is the target number itself a win or a loss? Each game defines this, and on a 0.00 to 99.99 range it shifts your chance by 0.01%. Read the rules panel once and you'll know.

The multiplier formula, worked through

The whole game turns on one line:

Multiplier = (100 - house edge) / win chance

Flip it around and you can find the edge from any payout the game shows you:

House edge = 100 - (multiplier x win chance)

So if a dice game offers 1.96x at 50% chance, the edge is 100 minus 98, which is 2%. If it offers 1.98x, it's 1%. That one check tells you more about a dice game's price than anything in its marketing.

Multiplier table at a 1% edge

Look at the last column. It never moves. Whatever target you pick, each dollar staked returns 99 cents on average at a 1% edge. The target changes how often you win and how big the wins are. It never changes the price.

What this means for a session

Turnover is what costs you. Expected loss equals total amount staked times the edge.

Drop $1 on 1,000 rolls and you've turned over $1,000. At 1% that's an expected cost of $10, whether you rolled at 98% chance or 2% chance. At 98% chance you'll see a smooth line drifting down. At 2% chance you'll see long flat losing runs broken by sudden 49.5x jumps. Same price, very different ride.

House edge: why 1% is common and why it varies

Many crypto casino originals set dice at about 1%. That makes it one of the cheapest games in the building. Slots typically carry 2% to 6%. Across all crypto casino games, few carry an edge that low.

But "common" isn't "guaranteed". Each operator sets its own edge, and some dice games run 2% or 3%. Studio-made dice games inside a slot lobby can differ again. Use the formula above on the game you're actually playing.

Comparing dice with its cousins

Dice sits alongside crypto plinko and Mines as an instant game with an operator-set edge. Plinko hides its edge inside a paytable you have to sum. Mines hides it in a multiplier ladder. Dice puts it in plain sight on every bet. If you like knowing exactly what you pay, dice is the most transparent of the three.

Provably fair: seeds, hashes and nonces

Provably fair is the reason crypto dice became popular in the first place. It lets you confirm the casino did not change your result after you placed your bet.

Three inputs decide each roll.

• Server seed: a secret random string from the casino. Before play, you see only its hash, a one-way fingerprint.
• Client seed: a string you set. Change it to anything you like.
• Nonce: a counter that starts at 0 or 1 and goes up by one with every bet on that seed pair.

The game combines them, runs them through a hash function such as HMAC-SHA256, and converts part of the output into a number between 0 and 99.99. Because the server seed was committed before you set your client seed, the casino can't pick a server seed that beats your bets. Because you can't see the server seed, you can't pick a client seed that beats the casino.

Why the nonce matters

Without a nonce, every bet on the same seed pair would produce the same roll. The nonce makes each bet unique while keeping the sequence reproducible. Bet number 37 always uses nonce 37 (or 36, depending on where the count starts). When you verify, you plug in the nonce from your bet history.

How to verify a roll

1. Open the fairness panel and copy the hashed server seed.
2. Set your own client seed.
3. Play. Each bet uses the next nonce.
4. Rotate your seed pair. The casino reveals the old server seed.
5. Hash the revealed seed with SHA-256 in any independent tool. It must match the hash from step 1.
6. Plug server seed, client seed and nonce into the casino's verifier, or an independent script, and compare each result with your history.

If step 5 fails, the casino swapped its seed after committing. That's a serious problem and worth raising with support, then the licensing authority.

What provably fair does not prove

It proves the roll wasn't altered. It doesn't prove the edge is low, the casino is solvent, or withdrawals will be paid. Those come from reading terms, which is what our Index measures.

Betting systems debunked: the Martingale maths

Dice is where betting systems go to die. The game is fast and the payouts are clean, so people try doubling schemes. The arithmetic says no.

How Martingale works

Bet $1 at roughly 50% chance (1.98x payout). If you win, start again at $1. If you lose, double the stake. When you eventually win, you recover all losses plus a small profit.

It feels unbeatable. Most sessions end in profit. That's the trap.

Where it breaks

At 49.5% win chance, your chance of losing any single roll is 50.5%. The chance of losing 10 in a row is 0.505 to the power of 10, about 0.11%. Rare, right?

Now count the bets. Over 1,000 rolls, the chance of hitting at least one run of 10 straight losses is about 41%. Over 5,000 rolls it's about 93%.

And what does 10 straight losses cost? $1 + $2 + $4 + ... + $512 = $1,023. Your next bet would be $1,024. Because 1.98x is slightly short of a true double, even a win there returns $1,003.52 of profit, leaving the sequence about $19.48 down. In fact, at 1.98x any sequence of six or more losses fails to break even when it finally wins. Most casinos also have a max bet that stops the sequence anyway.

The edge doesn't move

Every bet in a Martingale sequence still carries the same 1% edge. Total turnover goes up because stakes keep doubling, so expected loss in dollars goes up too. You've rearranged the shape of your losses into many small wins and one large wipeout. The price per dollar wagered is unchanged.

Other systems, same result

• Reverse Martingale (Paroli): double after wins instead. Smaller downside, rarely reaches the big win.
• D'Alembert: add one unit after a loss, subtract one after a win. Slower drift, same edge.
• Fibonacci: stakes follow the sequence 1, 1, 2, 3, 5, 8. Slower climb than Martingale, same eventual wall.
• Target switching: changing from under 50 to over 50 after a loss. The roll doesn't know which side you're on.

No staking pattern changes the expected return of independent bets. You can only change variance and how long a budget lasts.

Bonus wagering and crypto dice

Dice is a terrible way to clear a welcome bonus. Casinos know dice is low-edge and lets you place near-certain bets, so they cut its contribution hard.

Our brand data shows two clear examples. Bets.io counts slots at 100% and other games at 5%. Betpanda counts slots at 100%, live games at 20% and other games at 0%. Dice is "other games" at both.

A worked example

Take a $100 bonus at 40x on the bonus: $4,000 of turnover needed. On a slot with a 2% edge, expected cost is about $80. That's our Clear benchmark.

On dice at 5% contribution, each $1 rolled counts as $0.05. You'd need $80,000 of dice turnover. At a 1% edge that's around $800 of expected cost, ten times the slot route. At 0% contribution, as at Betpanda, it never clears at all.

Bonus terms also often ban "low-risk" play, such as betting both sides or rolling at 98% chance, as bonus abuse. That can void winnings. The Bets.io review notes a 2 USDT max bet while a bonus is live and broad terms around other games, so stay on slots until the bonus is cleared or gone.

The cleaner route

Play dice with cash only. If you want something back on losses, wager-free cashback is a better match for low-edge games. Cocobet states 50% first-week cashback on net losses, paid wager-free, and Winna offers instant rakeback. Our comparison of crypto casino cashback runs the numbers on each.

Choosing a coin and network for dice

Dice stakes can be fractions of a cent. Your deposit and withdrawal fees can easily cost more than your expected dice loss for the session, so coin choice matters.

• Stablecoins (USDT, USDC) keep your balance steady. You see your dice result, not a price chart.
• TRC-20 USDT usually has low network fees. Check what the cashier supports.
• Litecoin is a common low-fee option for withdrawals.
• Bitcoin on-chain fees vary with mempool congestion. Lightning cuts them; BC.Game and Betpanda confirm Lightning in our data.

Our list of litecoin casinos shows which brands confirm LTC. Always send on the network the cashier names. A coin sent on the wrong chain can be lost.

Fast withdrawals after a lucky roll

A 99x hit on a long-shot roll is fun until you have to wait to withdraw. Stated payout speeds vary a lot. Cybet, Cocobet, Winna, TrustDice, CasinoCrypto.io and Crypto-Games.io state instant payouts. Jack states under 5 minutes. Bets.io states instant to 48 hours and publishes deposit-linked daily caps of 1,000 to 3,000 USDT.

Our ranking of fast payout crypto casinos compares all 17 on stated speed. These are the operators' stated speeds. Every casino on it passed the Haxies group's multi-source testing and Team Lead Harrison Ayla's interview before being listed. A first withdrawal can also trigger a verification check even where signup needed no documents, so allow for that.

Strategy that actually helps

You can't beat dice. You can play it sensibly.

Pick the target for the session you want

High win chance (75% to 95%) gives a long, smooth session with small wins. It's a good fit for a fixed budget and a set number of rolls.

Middle targets (40% to 60%) give coin-flip swings. A 50-roll session will usually stay within a few stakes of break-even, minus about 1% of turnover.

Long shots (under 10%) give long losing runs and occasional big jumps. Budget for droughts. At 2% chance, going 150 rolls without a win happens about 5% of the time.

Control turnover

Because the edge is fixed per dollar, the real lever on cost is how much you stake in total. Fewer rolls, smaller stakes, slower autobet. That's it.

Use autobet stops

Every autobet panel has stop-on-loss and stop-on-profit fields. Fill them in before you press start. A stop-on-loss equal to your session budget is the most useful setting in the whole game.

Dice has two close relatives worth knowing. Crypto limbo uses the same win chance formula with the target set as a multiplier instead of a number to roll under, and crypto hi-lo applies the same pricing card by card, so the payout for each guess depends on the card in front of you.

Mistakes to avoid

• Not checking the edge. One multiplication tells you the price. Do it once per game.
• Running Martingale with no ceiling. Decide your maximum sequence length in advance, or better, don't run it.
• Playing with a bonus active. Low contribution and abuse clauses make it a losing deal.
• Fast autobet with no stops. Hundreds of rolls a minute burn through turnover very quickly.
• Treating a new client seed as a fresh start. It's a new sequence, not a better one.
• Sending coins on the wrong network. Confirm chain and address before every deposit.

Is crypto dice legal for you?

It depends on where you live, and sometimes on your region within a country. Each casino also publishes a restricted-country list. Check both before signing up.

Licences vary across our 17 brands: several publish Anjouan or Curaçao licence numbers, Winna lists the Tobique Gaming Commission, and Betpanda publishes no gambling licence, only a Costa Rica company registration. If privacy matters to you, our list of no KYC crypto casinos spells out what each operator says about verification at signup and at withdrawal.

Keeping dice in proportion

Dice is quick. On autobet it can run hundreds of rolls a minute, and crypto deposits arrive without bank friction. Set deposit limits, use cooling-off tools, and take breaks away from the screen. Good habits for responsible crypto gambling start before the first deposit, and Gambling Therapy offers free, confidential help in many languages.

Crypto dice questions

Is crypto dice rigged?

At a provably fair dice game, you can verify each roll by checking the revealed server seed against the hash shown before play and recomputing results with your client seed and nonce. That proves rolls weren't changed. Check the house edge separately with the multiplier formula.

What is the house edge on crypto dice?

It's set by the operator, and many in-house dice games run at about 1%. Find it yourself: multiply the payout by the win chance and subtract from 100. A 1.98x payout at 50% chance means a 1% edge.

How is the dice multiplier calculated?

Multiplier equals (100 minus the house edge) divided by your win chance. At a 1% edge, a 25% chance pays 3.96x and a 10% chance pays 9.9x.

Does Martingale work on dice?

No. It turns many small wins into rare large losses, and each bet still carries the same edge. At 49.5% chance, the odds of a 10-loss streak somewhere in 1,000 rolls are about 41%, and that streak costs 1,023 units.

Is rolling over better than rolling under?

No. Roll over 75 and roll under 25 have the same 25% chance and the same payout. Direction is a display choice, not an advantage.

What is a nonce in crypto dice?

A nonce is a counter that goes up by one with each bet on the same seed pair. It makes every roll unique while keeping the whole sequence reproducible for verification.

Can I clear a casino bonus on dice?

Usually not efficiently. In our brand data, Bets.io counts other games at 5% and Betpanda at 0%, and many terms treat low-risk dice betting as bonus abuse. Clear bonuses on slots and play dice with cash.

Which win chance should I pick?

Pick by the session you want. High chances give smooth, slow results; low chances give long droughts and big wins. The expected cost per dollar staked is the same at every target.