Compare

Casino dice games: understand the roll and its payout

Casino dice games can ask you to predict a number on a slider or follow a Craps wager through several rolls. In Shuffle Originals Dice, a result must land strictly above or below your chosen target. In Craps, the same total can mean something different before and after a point is established. Start with the winning condition on the actual game, then read its payout.

Which dice game are you looking at?

A target and an Over or Under control

Shuffle Originals Dice uses a digital scale from 0.00 to 100.00, in steps of 0.01. You choose a target, a direction and a stake. A correct prediction wins at the selected multiplier; an incorrect prediction loses that round's stake. The scale has 10,001 possible displayed values. It is a numbered result grid, so counting the displayed values alone does not tell you the probability of each one.

A layout with Pass, Don't Pass and a point

Craps uses the total of two dice. A Pass Line wager can settle on the come-out roll or stay active while the shooter tries to repeat an established point before rolling 7. The examples here use the illustrative rules in the MGM Grand Detroit Gaming Handbook. They explain the line wagers and its Field paytable; an individual live or digital table can have different side bets, payouts and limits.

The presentation does not decide the rules. A digital Craps game still needs its Craps settlement rules. A seed-verifiable threshold game needs its own number conversion and verification method. A game's name, a cryptocurrency balance or an animated pair of dice does not supply those details.

Shuffle Dice: set the condition before the roll

The stake, target, displayed win chance and multiplier belong to one wager. Shuffle describes linked controls: changing the slider or target updates the displayed chance and payout. A larger multiplier comes with a smaller displayed chance of winning. Read the final set of values together before submitting the wager.

  1. Choose the denomination and enter the stake for the next round.
  2. Select Under or Over. Under requires a result below the target; Over requires a result above it.
  3. Adjust the target, displayed chance or multiplier, then read the linked values. At a displayed 49.5% chance, the published nominal payout formula gives a 2x multiplier.
  4. Check the stake and the exact winning boundary. Manual mode submits one wager per action. Keep the round's chosen inputs if you intend to check its result later.

Equality loses at the threshold

Under 49.50: equality losesAdjacent displayed results: 49.49 wins, 49.50 loses and 49.51 loses. The rule illustration is not a probability chart. Under 49.50 49.4949.5049.51 WinLoseLoseResult must be strictly below
A matching target fails the Under condition. Nearby results are shown at illustrative spacing.

Under 49.50 wins on 49.49 and loses on 49.50 or 49.51. Over 50.50 loses on 50.50 and wins on 50.51. The word strictly matters: landing exactly on the target does not create a push under these published rules.

Under 49.50

Result
49.49
Settlement
Win

Under 49.50

Result
49.50
Settlement
Lose the stake

Under 49.50

Result
49.51
Settlement
Lose the stake

Over 50.50

Result
50.50
Settlement
Lose the stake

Over 50.50

Result
50.51
Settlement
Win

The boundary illustration shows adjacent displayed values around Under 49.50. The spacing is an explanation of the rule, with no claim about how often those values occur.

A 2x return includes your original stake

For a winning threshold wager, multiply the stake by the payout multiplier to find the gross return. Subtract the stake to find net profit. With stake s and multiplier m, gross return is s × m and net profit is s × (m − 1). A losing wager returns zero and loses the stake.

Hypothetical wager: 0.001 BTC, Under 49.50, 2x

Shuffle's published test inputs produce the result 43.05. Applying that number to this hypothetical wager, 43.05 is below 49.50, so the stake wins. The BTC amount is chosen for arithmetic; this is not a record of a played round.

Result 43.05, a win

Gross return
0.002 BTC
Net result
+0.001 BTC

Result exactly 49.50, a loss

Gross return
0 BTC
Net result
-0.001 BTC

Calling the 0.002 BTC return profit would count the original 0.001 BTC twice. The distinction also matters when you compare a multiplier with a table payout. A 2:1 profit payout earns twice the stake and returns three stakes in total. A 2x gross multiplier returns two stakes in total and earns one stake of profit.

Manual, Auto and Advanced change the stake sequence

Manual

Shuffle describes one action for each roll, followed by reviewing the outcome and deciding whether to change the next stake or target. The most direct accounting is one stake, one saved condition and one settlement.

Auto

Auto runs a series with a bet count or stopping conditions. Shuffle describes stop-loss and stop-win controls. Before starting a series, read the selected count and stop settings alongside the stake. Those controls schedule wagers; a stop setting should not be treated as proof that every possible loss will end at exactly that displayed amount.

Advanced

Advanced uses presets or custom rules to change stakes after wins and losses. Shuffle names Martingale, Delayed Martingale, Paroli and D'Alembert, and provides a Create Strategy option. A preset changes how much subsequent wagers risk. It does not improve the underlying odds or make recovery of earlier losses certain.

Result history describes completed outcomes. It can help reconcile which wagers settled, but a losing run is not a reason that the next roll must win. For a changing-stake sequence, count the actual stakes placed, including increases. Multiplying the first stake by the number of rounds can understate the total exposure.

Shuffle publishes a displayed chance floor of 0.01% and a maximum nominal multiplier of 9,900x. That is a multiplier limit, not a fixed cash award. The maximum stake varies with the target multiplier. Use the selected denomination's displayed stake limits and settlement precision for an actual wager; a wallet's currency label alone does not establish the unit of a separate limit formula.

What Shuffle's published 99% RTP describes

Shuffle publishes a nominal 99% RTP and 1.00% house edge, with multiplier m = 0.99 ÷ p. Here p is the displayed chance expressed as a fraction: 49.5% becomes 0.495. The nominal calculation is 0.99 ÷ 0.495 = 2. Feeding the number 49.5 into that formula would use the wrong unit.

At p = 0.495 and m = 2, expected gross return per stake is 0.495 × 2 = 0.99 times that stake. Expected net is -0.01 times the stake. The denominator is the sum of all wager stakes, including money staked again after a win. It is not the deposit, the ending balance or only the stakes of winning rounds.

Hypothetical turnover calculation

Suppose 1,000 wagers each have a stake of 0.001 BTC. Their cumulative stakes total 1 BTC. The nominal model gives an expected gross return of 0.99 BTC and expected net result of -0.01 BTC across that turnover. This arithmetic is neither a proposed sequence of wagers nor a prediction of a session result. A particular series can finish above or below its expectation.

A 99% RTP also says nothing about winning 99% of rounds. In this nominal example, the displayed win chance is 49.5%; the return calculation combines that chance with a 2x gross payout.

Why the exact result grid needs its own calculation

Shuffle's published Dice conversion uses the first four HMAC bytes to form a number, then floors that number onto the hundredth grid. For this documented case, with one result and cursor 0:

  1. Treat the four bytes as a big-endian integer n, from 0 to 4,294,967,295.
  2. Divide n by 4,294,967,296 to obtain the float f.
  3. Calculate floor(f × 10,001) ÷ 100 to obtain the displayed Dice result.

Floor means round down to the integer at that step. Rounding the float to the nearest hundredth is a different procedure and can produce a different result.

In an ideal mathematical model where every four-byte string is equally likely, there are 2³² possible inputs. They cannot be divided evenly among 10,001 display bins. Under 49.50 accepts grid indices below 4,950, so its winning input count is ceil(4,294,967,296 × 4,950 ÷ 10,001) = 2,125,796,232. The resulting probability is 2,125,796,232 ÷ 4,294,967,296, approximately 0.4949505049735.

If that conditional probability is paired with a hypothetical 2x gross payout, expected gross return is approximately 0.9899010099471 times the original stake. Expected loss is approximately 1.0098990053% of that stake. Keep this result separate from Shuffle's published nominal 1.00% statement: the calculation assumes uniformly distributed four-byte inputs and a 2x payout, and it does not establish a currently served engine's return or coin rounding. It provides no betting advantage or evidence that a game was manipulated.

Check a completed Shuffle result with its old inputs

Shuffle describes a commitment with a secret server seed, a player client seed and a nonce. The server seed is represented as a random 64-character hex string; its SHA-256 commitment is displayed while the secret remains hidden. The client seed initially comes from the browser and can be changed by the player. The nonce starts at 0 and increases for new bets made with the same seed pair.

Keep the commitment from before the wager, the old client seed and nonce, the wager's target and multiplier, and the saved result. Once the old server seed is revealed, there are two checks: whether it matches the earlier commitment and whether it reproduces the result.

The published reveal procedure

  1. Complete existing games before changing the seed pair.
  2. Open the old bet's modal and choose Rotate pair.
  3. Enter the desired next client seed and choose Change. This commits a new pair for subsequent bets.
  4. Close the provably fair modal and return to the particular old bet. Its old server seed is now revealed; that bet's client seed and nonce remain unchanged.
  5. Choose Verify using the old client seed, revealed old server seed and old nonce. Use the Dice conversion for that implementation, then compare the number with the saved result.

For the commitment comparison, use the matching implementation's seed representation and verifier. A 64-character hex string can be treated as text or decoded into bytes; those choices are not interchangeable. Compare the calculated SHA-256 with the hash recorded before the wager. Without that earlier hash, reproducing a number alone does not show that the seed was committed in advance.

Published test vector: the number 43.05

The known example uses client seed abc, revealed server seed def, nonce 1 and cursor 0. Its HMAC-SHA256 message is abc:1:0, keyed by def. The resulting hash is 6e37d85daa228fdf2caeabc192b01502d99bc5e4433994bcabac23403bf407a5.

First four bytes

Value
110, 55, 216, 93

Big-endian integer

Value
1,849,153,629

Float

Value
1,849,153,629 ÷ 4,294,967,296 = 0.43053962965495884

Dice result

Value
floor(0.43053962965495884 × 10,001) ÷ 100 = 43.05

These are published test inputs, not the secret seed of a future round. For the hypothetical Under 49.50 wager, the reproduced 43.05 wins; settlement still needs its saved stake and multiplier. Verifying the generated number and checking its payment arithmetic are separate tasks.

Use the old pair throughout. A new seed pair or wrong nonce can produce a different result. If a required field is missing, the check is incomplete; that alone does not show unfairness. Matching a commitment and number does not predict a future roll or establish withdrawal reliability.

For a fuller commitment-and-replay walkthrough, continue to the provably fair explanation.

Verify a roll

Craps: the come-out roll and the point have different rules

Under the MGM Grand Detroit handbook's illustrative rules, the come-out roll begins the line wager. A total of 4, 5, 6, 8, 9 or 10 establishes a point. Once there is a point, the main line wager waits for that number or 7; the other totals keep it active.

Pass Line

On the come-out roll, 7 or 11 wins, and 2, 3 or 12 loses. If a point is established, repeating the point before 7 wins; rolling 7 before the point loses. An 11 during the point phase does not receive the come-out win treatment.

Don't Pass

On the come-out roll, 2 or 3 wins, 7 or 11 loses, and 12 pushes. After a point is established, 7 before the point wins; the point before 7 loses. The come-out 12 push is why Don't Pass is not a complete reversal of Pass Line.

Come-out 7 or 11

Pass Line
Win
Don't Pass
Lose

Come-out 2 or 3

Pass Line
Lose
Don't Pass
Win

Come-out 12

Pass Line
Lose
Don't Pass
Push

Come-out 4, 5, 6, 8, 9 or 10

Pass Line
Establish the point
Don't Pass
Establish the point

Point repeated before 7

Pass Line
Win
Don't Pass
Lose

7 before the point

Pass Line
Lose
Don't Pass
Win

Other total during the point phase

Pass Line
Wager continues
Don't Pass
Wager continues

Line wins pay even money. A 10 EUR winning line wager earns 10 EUR profit and returns 20 EUR including the stake. A loss returns zero. A push returns the original 10 EUR and earns zero profit. EUR is used here for arithmetic, with no claim about a particular table's currency or minimum.

Hypothetical round: point 6, then 11, then 6

Point 6 changes what the next roll meansHypothetical main line wager: come-out 6 establishes point 6. A later 11 continues the wager. Point 6 before 7 wins Pass Line and loses Don’t Pass; 7 first reverses those outcomes. Come-out: 2 + 4 = 6Point 6 is establishedNext roll: 5 + 6 = 11Both line wagers continue 6 before 7Pass winsDon’t Pass loses7 before 6Pass losesDon’t Pass winsAlternative endings, not extra bets
The example takes the 6-before-7 branch. A point-phase 11 does not settle the main line.
  1. The come-out dice show 2 + 4 = 6. Point 6 is established, so the line wager stays active.
  2. The next dice show 5 + 6 = 11. The point is already 6, so the line wager continues. No line payout occurs on this roll.
  3. The next dice show 3 + 3 = 6. The point repeats before 7. A 10 EUR Pass Line wager returns 20 EUR, including 10 EUR profit. A separately considered 10 EUR Don't Pass wager loses its 10 EUR stake.

If 7 had arrived before the repeated 6, Pass Line would lose and Don't Pass would win. If the come-out roll had instead been 12, Pass Line would lose and Don't Pass would return its stake as a push. These are alternative branches of the example, not payouts to add together.

Field and other Craps wagers settle on their own terms

The Field wager in the same handbook settles on the next roll, whether the main line is on the come-out roll or already has a point. Its payout belongs to the Field stake alone.

3, 4, 9, 10 or 11

Field profit payout
1:1
Gross return on a 10 EUR stake
20 EUR

2

Field profit payout
2:1
Gross return on a 10 EUR stake
30 EUR

12

Field profit payout
3:1
Gross return on a 10 EUR stake
40 EUR

5, 6, 7 or 8

Field profit payout
Lose
Gross return on a 10 EUR stake
0 EUR

A hypothetical 10 EUR Field wager on 12 earns 30 EUR profit and returns 40 EUR in total. A separately placed Don't Pass wager pushes on a come-out 12. The total on the dice is identical; the wager and its phase decide the settlement. Other tables may use a different Field payout for 12, so carry this paytable's name with its numbers.

Come and Don't Come

The handbook places these wagers after a point has been established. Come follows the corresponding Pass rules, while Don't Come follows Don't Pass. Each is a separate wager with its own sequence. An active Pass Line point does not turn every new Come wager into a bet on that existing point.

Odds behind a line or Come wager

Once a point is established, Odds is an additional wager supporting Pass Line, Come, Don't Pass or Don't Come. Its payout varies with the point number. Include that additional stake when counting exposure. A maximum Odds multiple and its complete payout schedule must come from the chosen table; the line-bet calculations here do not include an Odds stake.

Place, Buy and Lay

Place wagers back 4, 5, 6, 8, 9 or 10 before 7. In this handbook, Buy wagers use the same event as Place with a 5% commission and true odds. The Buy commission's calculation base is not specified there. Lay wagers back 7 before the selected number and charge 5% of what the bet will win. A commission percentage needs its base before you can calculate the monetary charge.

Hardways and optional wagers

Hardways target 4, 6, 8 or 10 in the exact double shown on the layout. The same total in another combination, or a 7, loses the wager. Its payout is a separate paytable. The handbook also describes Big 6 & 8 only when available and an optional Golden Dice Challenge, so those names should not be treated as a complete menu or a reward promise for another table.

Theoretical line-bet returns need a stated denominator

For independent, fair, standard six-sided dice, each roll has 36 ordered face pairs. Applying the line rules above gives the following conditional mathematical results. These values describe that model and those main wagers; they are not a measurement of a physical table or a return for all dice games.

Pass Line: include the entire wager through its point

The come-out wins immediately on 7 or 11, accounting for 8 of the 36 ordered pairs. A point can also be made before 7. Combining those point paths gives a win probability of 244/495 and loss probability of 251/495. With even-money profit, expected loss per original stake is 251/495 − 244/495 = 7/495, approximately 1.41414%.

Intermediate rolls which leave the point unresolved do not create another main line stake in this calculation. Counting them as new wagers would change the denominator.

Don't Pass: keep the come-out 12 push in the count

For the same fair-dice model, Don't Pass wins with probability 949/1980, loses with 976/1980, and pushes with 55/1980 = 1/36. A push returns the stake and contributes zero net profit. Expected loss is (976 − 949) ÷ 1980 = 27/1980 = 3/220, approximately 1.36364% of the initial stake, including wagers that push.

If you exclude pushes and count only decisive wins and losses, the corresponding ratio becomes 27/1925, approximately 1.402597%. It answers a different accounting question. Compare returns using the same denominator; do not replace an initial-stake percentage with a decisive-only percentage halfway through a comparison.

To compare these line wagers with other table games, use the table-game RTP explanation and keep each game's own rules beside its return.

Compare table rules

Match the next check to the game

For a Shuffle-style threshold result, keep the old commitment, seed pair, nonce, target and multiplier together and check the completed round. For Craps, identify the wager, whether it is the come-out or point phase, and the exact table paytable before calculating the return. Casino game guides provide the next explanation when you move to a different rule set.

Browse game guides

Lewis Crane

Lewis Crane

Games and slots writer

Games and slots writer covering mechanics, providers, RTP and volatility.

Last updated - By Lewis Crane