Four dice games, one called Classic Dice
"Dice at BC.Game" can mean four different BC Originals. They share a seed system and differ in range, top multiplier and verifier, so the first job is to know which one is open.
| Game | Result range | RTP in game data | Max win in game data | BC.Game verifier |
|---|---|---|---|---|
| Classic Dice | 0.00 to 100.00 | 98 | 9900× | classicdice_new |
| Dice | 0 to 9999 in the verifier (0.00 to 99.99 in the fairness text) | 98 | 9800× | dice |
| Hash Dice | 0 to 99999 | 98 | not listed | hashdice |
| Ultimate Dice | 0 to 9999 in the verifier | 98 | 9,900x | ultimatedice_new |
Game data on bc.game, 13 Nov 2026. Ranges from BC.Game's verifiers and help centre.
The rest of this page is about Classic Dice, the one most people mean. For Hash Dice the help centre gives its own figures: a result between 0 and 99,999, a win of "at least 1.0102x, and the maximum could go up to 99,000 x", and "only 1% edge" (article dated 28 November 2025). Those multipliers are the ones a 99% game produces, which the game data's 98 contradicts.
What BC.Game states about Classic Dice
| Item | What BC.Game states | Where |
|---|---|---|
| RTP in the game data | 98; the info panel showed a house edge of "2.00%" on 1 Nov 2026 | Game data on bc.game (13 Nov 2026) |
| House edge in the help article | "The Classic Dice game has a House Edge of only 1%" | Help centre, How to Play Classic Dice, dated 28 November 2025 (13 Nov 2026) |
| Originals in general | "We recently adjusted the Return to Player (RTP) percentage for our original games, changing it from 99% to 96%" | Help centre, House Edge/RTP, dated 26 November 2025 (13 Nov 2026) |
| Winning rolls | "only roll outcomes that hit the green area are winners" | Help centre (13 Nov 2026) |
| Scripts | "the use of scripts is optional. Players assume full responsibility for any attendant risks" | Help centre (13 Nov 2026) |
| Auto mode | On win / on lose: increase by a set amount or reset; stop on win; stop on loss | Help centre (13 Nov 2026) |
The three price statements cannot all be true, and for dice there is a quick way to see which applies: the multiplier the game offers at a 50% win chance. A game priced at 99% offers 1.98x there, one at 98% offers 1.96x and one at 96% offers 1.92x. The game client on 1 Nov 2026 computed the multiplier as 98 divided by the win chance in percent, rounded down to four decimals, which is the 98% version.
The price, target by target
A Classic Dice roll is one of 10,001 values, 0.00 to 100.00 in steps of 0.01. A bet at a win chance of c% covers c × 100 of those values. The return is the multiplier times that fraction.
| Win chance | Multiplier at 98 | Winning rolls | Return | Multiplier a 99% game would show | Multiplier a 96% game would show |
|---|---|---|---|---|---|
| 0.01% | 9,800x | 1 of 10,001 | 97.99% | 9,900x | 9,600x |
| 1% | 98x | 100 of 10,001 | 97.99% | 99x | 96x |
| 10% | 9.8x | 1,000 of 10,001 | 97.99% | 9.9x | 9.6x |
| 25% | 3.92x | 2,500 of 10,001 | 97.99% | 3.96x | 3.84x |
| 50% | 1.96x | 5,000 of 10,001 | 97.99% | 1.98x | 1.92x |
| 75% | 1.3066x | 7,500 of 10,001 | 97.98% | 1.32x | 1.28x |
| 90% | 1.0888x | 9,000 of 10,001 | 97.98% | 1.1x | 1.0666x |
| 98% | 1x | 9,800 of 10,001 | 97.99% | 1.0102x | under 1x |
Multiplier rule: BC.Game Classic Dice game client (1 Nov 2026). Returns are arithmetic over the 10,001 possible rolls.
Every target returns between 97.98% and 97.99%. The hundredth of a percent below 98 comes from the 10,001st value: the multiplier is set as if there were 10,000 outcomes. The spread between targets is rounding and not something to exploit.
The top of the range is worth a second look. The lowest win chance the game accepted on 1 Nov 2026 was 0.01%, where 98 / 0.01 gives 9,800x. The game data lists a maximum win of "9900×" for Classic Dice, a figure that belongs to a 99% game. If the panel offers 9,900x at 0.01%, the game is being priced at 99% at that moment; if it offers 9,800x, the listed maximum is out of date.
How a roll is derived
BC.Game publishes two descriptions of the calculation, and they are not the same. Its fairness text says: compute HMAC_SHA256(clientSeed:nonce, serverSeed), "take 8 characters of the hash and convert it to an int32 value", divide by 0x100000000, "multiply it by 10001 and divide it by 100". The help article describes the same eight-character step.
The verifier BC.Game links from the game, classicdice_new, does something else with the same hash:
hash = HMAC-SHA256(key = server seed, message = clientSeed:nonce)
index = (first 15 hex characters of hash) mod 47
n = the 14 hex characters starting at index, shifted right by 3 bits (53 bits)
roll = n × 2⁻⁵³ × 10001 / 100
Both methods spread results evenly over the 10,001 values, so the price is unaffected. They do not give the same roll for the same seeds, which matters when you check one. The page the game links is the one to trust for a live bet.
Our dice verifier has a BC.Game mode that follows the eight-character method of the fairness text. It will confirm that the revealed server seed matches the hash you were shown, which is the commitment check and holds for either method, but its roll can differ from the sliding-window verifier's. For the roll itself use BC.Game's classicdice_new page. As with every seed-based BC Original, the server seed is revealed only after you set a new one.
Classic Dice against dice elsewhere
| BC.Game Classic Dice | Stake | Shuffle | Roobet | |
|---|---|---|---|---|
| Stated price | 98 in the game data; 1% in the help article | 99% RTP | 1.00% house edge | 99% RTP |
| Multiplier at a 50% win chance | 1.96x | 1.98x | 1.98x | 1.98x |
| Possible rolls | 10,001 | 10,001 | 10,001 | 10,000 (0.00 to 99.99) |
| Return by formula | 97.99% | 98.990% | 98.990% | 99% |
| Top multiplier | 9,800x at 0.01% | 9,900x | 9,900x | Shown in the game |
| Hash | HMAC-SHA256, 53-bit window | HMAC-SHA256, first 4 bytes | HMAC-SHA256, first 4 bytes | HMAC-SHA512 |
Stake, Shuffle and Roobet: each casino's documents as of 12 Nov 2026, detailed on the pages linked below.
The practical difference is the two cents: BC.Game pays 1.96 where the other three pay 1.98 on an even-money roll. Over $1,000 of bets that is about $20 of expected cost against about $10.
Streaks at each target
Because the price is the same at every target, the useful question about a target is how long it can go without paying. The chance of losing n rolls in a row is the losing chance raised to the power n.
| Win chance | Multiplier | 5 losses in a row | 20 losses in a row | 100 losses in a row | Rolls for an even chance of one win |
|---|---|---|---|---|---|
| 90% | 1.0888x | 1 in 100,000 | under 1 in a billion | under 1 in a billion | 1 |
| 50% | 1.96x | 3.1% | 1 in 1,000,000 | under 1 in a billion | 2 |
| 25% | 3.92x | 23.7% | 0.3% | under 1 in a billion | 3 |
| 10% | 9.8x | 59.1% | 12.2% | 1 in 38,000 | 7 |
| 2% | 49x | 90.4% | 66.8% | 13.3% | 35 |
| 1% | 98x | 95.1% | 81.8% | 36.6% | 69 |
Arithmetic over the 10,001 possible rolls. Each roll is independent of the ones before it.
At a 1% target, 100 losing rolls in a row happens about 37% of the time from any starting point, and a run of 20 about 82% of the time. At 50%, twenty losses in a row is about one in a million. Neither run changes what the next roll is worth.
Hash Dice has its own calculation
Hash Dice does not share Classic Dice's method. The help centre describes it as: combine server seed, client seed and nonce, take a SHA-512 hash, read "the first 5 characters", and if the value is below one million use it modulo 100,000; otherwise "it repeats using the next five characters up to 25 times". The verifier BC.Game links for Hash Dice applies that five-character rule to HMAC-SHA512 of clientSeed:nonce keyed with the server seed. The game offers "High" and "Low" intervals on a 0 to 99,999 scale. The 1.0102x minimum and 99,000x maximum in the article are 99 divided by a 98% and a 0.001% win chance; under the 98 in the game data the same two targets would pay 1.00x and 98,000x.
What the slider changes
Moving the target trades frequency for size at a constant price. At 90% you win nine rolls in ten and gain about 8.9% of the stake each time; at 1% you win one roll in a hundred and gain 97 times the stake. The expected loss per dollar staked is the same two cents in both cases. Roll-over and roll-under are mirror images, and neither is favoured.
The auto mode's ON LOSE setting, which can raise the next bet after a loss, is a Martingale by another name. It raises the chance of ending a short session ahead by a small amount and pays for that with a rare, large loss; the expected result per dollar wagered stays at minus two cents because each roll is computed from the seeds and the nonce alone. The Martingale simulator and the dice calculator show the numbers. Account terms are in the BC.Game review.
Sources: BC.Game help centre, How to Play Classic Dice, How to Play Hash Dice and House Edge/RTP (13 Nov 2026); game data on bc.game (13 Nov 2026); Classic Dice fairness text, info panel and game client on bc.game (1 Nov 2026); BC.Game Classic Dice verifier (13 Nov 2026). Returns are our arithmetic.