Dice odds: win chance, multiplier and losing streaks
Crypto dice picks a number from 0 to 100, and you decide how much of that range counts as a win. Divide 99 by your win chance to get the multiplier: 49.5% pays x2, 10% pays x9.9. Multiply any pair back together on a 99% game and you're back at 99.
Win chance and multiplier
| Win chance | Multiplier | Roll over | Roll under | Wins about 1 in |
|---|---|---|---|---|
| 98% | x1.0102 | 2.00 | 98.00 | 1.02 |
| 90% | x1.1000 | 10.00 | 90.00 | 1.1 |
| 75% | x1.3200 | 25.00 | 75.00 | 1.3 |
| 66% | x1.5000 | 34.00 | 66.00 | 1.5 |
| 50% | x1.9800 | 50.00 | 50.00 | 2 |
| 49.5% | x2.0000 | 50.50 | 49.50 | 2 |
| 33% | x3.0000 | 67.00 | 33.00 | 3 |
| 25% | x3.9600 | 75.00 | 25.00 | 4 |
| 10% | x9.9000 | 90.00 | 10.00 | 10 |
| 2% | x49.5000 | 98.00 | 2.00 | 50 |
| 1% | x99.0000 | 99.00 | 1.00 | 100 |
| 0.1% | x990.0000 | 99.90 | 0.10 | 1,000 |
| 0.01% | x9,900.0000 | 99.99 | 0.01 | 10,000 |
Own calculation, multiplier = 99 ÷ win chance in percent. Stake's game pages give 9,900x as the top Dice multiplier, which matches a 0.01% win chance.
Pick two rows of any site's dice table and multiply them. 49.5 times 2 gives 99, 10 times 9.9 gives 99. A site where the product is 98 or 97 keeps two or three times as much, and its rules ought to say that somewhere. You can type any chance into the dice calculator and see the multiplier it should pay.
How the roll is made
Stake's provably fair notes put the roll somewhere from 00.00 to 100.00, 10,001 possible results, made from one float built out of both seeds and the nonce. Over and under mirror each other. 49.5% under 49.50 and 49.5% over 50.50 are the same bet with the slider on the other side.
How long the losing runs get
| Win chance | Losing run | Within | Chance it happens |
|---|---|---|---|
| 49.5% | 5 in a row | 100 rolls | 82.38% |
| 49.5% | 8 in a row | 1,000 rolls | 87.99% |
| 49.5% | 10 in a row | 1,000 rolls | 41.30% |
| 49.5% | 12 in a row | 1,000 rolls | 12.63% |
| 90% | 2 in a row | 100 rolls | 59.85% |
| 90% | 3 in a row | 1,000 rolls | 59.39% |
| 90% | 4 in a row | 1,000 rolls | 8.59% |
| 98% | 2 in a row | 1,000 rolls | 32.43% |
| 98% | 3 in a row | 1,000 rolls | 0.78% |
Own calculation, exact, by tracking the length of the current losing run roll by roll. Script: growth6/daten/phase2/b1a/dice.py.
On-loss settings in autobet are where this bites. Eight losses in a row show up somewhere in 1,000 rolls at 49.5% in 88 sessions out of 100. With double on loss switched on, bet nine is then 256 times your base bet, on a roll that still only wins 49.5% of the time.
The 98% setting goes wrong the other way. Nearly every roll wins and each win adds about one percent, then a single loss takes back around a hundred of them. Two losses back to back happen in roughly one session in three at that setting.
What 1,000 rolls cost
- Per $1 roll
- 1 cent on average
- Per 1,000 rolls of $1
- $10 on average, whatever win chance you choose
- Per $100 roll
- $1 on average
Nobody's session ends on exactly $10 down. At 49.5%, after 1,000 rolls you're usually somewhere between $42 down and $22 up. At 1% the swings run to hundreds of dollars, and ten hits, the number you'd expect, still leave you $10 short.
Dice strategies and predictors
Paroli, Martingale, d'Alembert and the "low risk" 98% grind all move money around inside a session and still give back 99% per dollar bet. Predictor bots are another matter. The server seed for your rolls only shows up after you swap it out, so a bot can't read it ahead of time and is after your deposit or your account. Past rolls you can check: switch to a new seed pair, then use a verifier.
18+. Dice has a built-in house edge at every win chance. Set a budget and a stop on loss before using autobet. Help: Responsible gaming. Not affiliated with Stake. Names belong to their owners.
Questions
How is the dice multiplier calculated?
On a 99% game, divide 99 by the win chance in percent. A 25% chance pays 99 ÷ 25 = x3.96.
How big is the edge on Stake Dice?
Stake states 99% for Dice, the same at every win chance, leaving the house 1%.
Which win chance is best in dice?
None is better on average. Every setting returns the RTP. High win chances give frequent small wins and rare heavy losses, low ones the reverse.
Is there a low risk dice strategy?
A high win chance makes losses rarer, not cheaper on average. At 98%, one loss wipes out roughly a hundred wins.
Do dice predictors work?
No. A roll depends on a server seed that remains hashed until you rotate. Predictor bots and scripts are a scam, whatever they promise.
Sources
- Stake: Dice game page (Wayback, 24 Oct 2026) · current page on stake.com. 99% RTP, roll over and roll under. Live page returns 403 to our fetcher. Checked 12 November 2026
- Stake: provably fair game events (Wayback, 19 Jul 2025). roll spread 00.00 to 100.00, 10,001 possible outcomes, one float per roll Checked 12 November 2026
18+ · Play responsibly · gamblingtherapy.org