September 2, 2026
Stake Dice Bot: Config Knobs and Risk Profile
A dice bot on Stake.com places bets mechanically using the same four-step loop every casino bot runs: send the bet amount and win condition, receive the roll result, adjust the next bet per your progression, and repeat or stop. The bot interacts with two parameters per round: a win chance from 0.01% to 98% and a bet amount. The payout multiplier is computed as 99 divided by your win chance, so 49.5% pays roughly 1.98x and 2% pays roughly 49.5x. That 99 instead of 100 is where the 1% house edge lives, and no bot, strategy, or progression removes it. What the bot does give you is disciplined execution of parameters you configured while calm, not at 3 AM when the losing streak hits.
This is the pillar guide for everything we write about dice bot configuration at StakeProSoft. It covers how the bot places each bet against Stake’s provably fair system, which progression strategies are available and how they differ, how payout and win chance interact to shape your risk profile, and the bankroll protection layers that keep a session from wiping your balance. For the broader framework of what automation does and does not do across all game types, see our casino bot buying guide.
How a Dice Bot Places Bets on Stake
A dice bot sends two values per round to Stake’s API: the bet amount and the win condition, which is your roll-over or roll-under target mapping to a win chance between 0.01% and 98%. The casino returns the roll result (a number from 0.00 to 99.99), the nonce for that bet, and the win or loss outcome. The bot reads the outcome, passes it to the progression logic, sizes the next bet, and repeats. That is the entire mechanical loop. For the full breakdown of how this loop works across all Stake games, see our Stake bot pillar guide.
Provably Fair: What the Bot Interacts With
Each dice roll is determined by HMAC-SHA256 using four inputs: a server seed generated by Stake and shown to you only as a SHA-256 hash commitment before the round, your client seed, an incrementing nonce, and a cursor fixed at 0 because dice needs only 4 bytes of hash output. The hash output is converted to a float between 0 and 1, then scaled to the 0.00 to 100.00 range via floor(float x 10001) / 100, producing 10,001 equally likely outcomes. Your win chance is the fraction of those outcomes that fall inside your winning zone.
The bot does not interact with seed generation. It cannot. The server seed is cryptographically locked before the session starts and shown only as a hash. No software can reverse it, pre-read it, or influence it. A bot that logs the server seed hash, client seed, and nonce per bet gives you a verification trail you can check after the session using Stake’s fairness settings. A bot that does not log these values leaves you with no way to audit the run.
The Payout Formula and House Edge
The payout multiplier is computed as 99 divided by your win chance. At 49.5% win chance, the multiplier is 99 / 49.5 = 1.98x. At 10% win chance, it is 99 / 10 = 9.9x. At 2%, it is 99 / 2 = 49.5x. The maximum manual payout is 9900x at the 0.01% win chance setting, since 99 / 0.01 = 9900. Stake’s own dice page publishes the full multiplier table and confirms the 1.00% house edge.
The 1% house edge lives in that 99, not in the roll. A fair coin flip at 50% win chance would pay 2.0x. Stake pays 1.98x. The expected value per dollar staked is (win chance x multiplier) minus 1, which equals (0.495 x 1.98) minus 1, which equals negative 0.01. That negative 1% applies to every bet the bot places, at every win chance, regardless of what your progression does. The progression changes the distribution of wins and losses across the session. It does not change the long-run average.
What the Bot Can and Cannot Do
| What a Dice Bot Can Do | What a Dice Bot Cannot Do |
|---|---|
| Place bets faster and more consistently than a human | Predict the next RNG outcome |
| Apply a progression without emotional deviation | Remove or reduce the 1% house edge |
| Enforce stop-loss and profit-target rules automatically | Guarantee a profit over any session |
| Run 24/7 without fatigue or distraction | Reverse or pre-read the server seed |
| Log every bet for post-session verification | Change the mathematical odds of any game |
The right column matters as much as the left. Any seller claiming their bot predicts outcomes or “cracks the seed” is selling fiction. The seed is cryptographically secured before the session. A bot places the bet and reads the result exactly as you would manually. What provably fair does give you is auditability: every bet the bot places can be independently verified after the session. But that is verification, not prediction.
Strategy Modes: Martingale, Fibonacci, and Alternatives
A progression strategy is the rule that determines how the bot sizes the next bet after a win or a loss. The progression changes the variance of your session, not the expected value. Every bet at every win chance has a negative expected value of 1% of the amount wagered. What changes is how wins and losses are distributed across the session and how quickly a losing streak can compound against your bankroll. The comparison below models each mode on a 100 EUR bankroll with a 0.10 EUR base bet at 49.5% win chance, the Stake default.
Flat Betting
Flat betting is the simplest mode: the bot places the same bet amount every round regardless of the previous outcome. There is no progression. A 0.10 EUR bet at 49.5% win chance loses 0.001 EUR on average per roll, and 1000 rolls lose roughly 1.00 EUR on average. Flat betting has the lowest variance and the lowest bust probability of any mode because the bet size never escalates. It is the baseline against which every other progression is measured. If your goal is to grind through wagering requirements with minimal risk of a sudden wipe, flat betting is the mode that will not surprise you.
Martingale
Martingale doubles the bet after each loss and resets to the base bet after a win. The logic is that a single win at any depth recovers every previous loss in the streak plus one base bet of profit. After 3 consecutive losses at 0.10 EUR base, you have lost 0.70 EUR. Your 4th bet is 0.80 EUR. If it wins, you gain 0.80, covering the 0.70 loss and netting 0.10 profit. Then you reset to 0.10.
The cost is exponential bet growth. Each loss doubles the stake. By loss 7, your bet is 6.40 EUR. By loss 9, it is 25.60 EUR. The 10th bet would be 51.20 EUR, but the remaining balance is only 48.90 EUR. The progression busts at loss 10 on this bankroll, not because the strategy logic fails but because the bankroll runs out. In a 1000-bet session at 49.5% win chance, the probability of encountering a 10-loss streak is roughly 67%. For the full survival math, including modeled drawdown at each loss depth and recovery probability tables, see our Martingale vs Fibonacci comparison.
Fibonacci
The Fibonacci progression uses the sequence 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233, 377, 610. Each number is the sum of the previous two. Your bet size is the current sequence number multiplied by your base unit. On a loss, you move one step forward. On a win, you move two steps back. If you are at step 1 or 2 and win, you reset to the base bet.
The bet sizes grow by a factor of roughly 1.618 per step (the golden ratio), not by 2. That slower growth is the point: you survive longer because each step costs less than a Martingale doubling. On the same 100 EUR bankroll with 0.10 EUR base, Fibonacci survives 14 consecutive losses before busting at loss 15, when the 15th bet of 61.00 EUR exceeds the remaining 1.40 EUR. In a 1000-bet session, the probability of hitting a 15-loss streak is roughly 9%. The tradeoff is recovery. A single Fibonacci win does not clear the slate. After 7 losses, full recovery takes 4 consecutive wins. After 9 losses, it takes 5. Each loss during the recovery phase pushes you one step deeper, extending the path back.
Custom Multipliers and Other Progressions
Some bots support custom multiplier sequences or alternative progressions like D’Alembert (increase by one unit on loss, decrease by one unit on win) or Labouchere (a cancellation system based on a target profit sequence). These sit between Martingale and Fibonacci in both aggression and recovery complexity. The principle is the same: the progression changes how bets are sized across the session, not the expected value per bet. If a bot exposes an editable multiplier list, you can define a custom sequence and test it at your chosen win chance before committing real funds. If the bot hides its parameters behind an “AI optimizer” with no visible configuration, you cannot verify what it does, and you should not spend money on it.
Strategy Mode Comparison
| Mode | How It Adjusts Bets | Bust Depth (100 EUR, 0.10 base, 49.5%) | Recovery | Best For |
|---|---|---|---|---|
| Flat | Same bet every round | Never escalates | N/A | Lowest variance, baseline, wagering |
| Martingale | Double on loss, reset on win | 9 losses (10th bet busts) | One win clears everything | Fast recovery, needs large bankroll |
| Fibonacci | +1 step on loss, -2 steps on win | 14 losses (15th bet busts) | Multiple consecutive wins needed | Longer survival, gradual recovery |
| D’Alembert | +1 unit on loss, -1 unit on win | Very deep (linear growth) | Gradual, one win at a time | Conservative progression, low variance |
| Custom | Editable multiplier list | Depends on sequence | Depends on sequence | Tinkerers who want full control |
The bust depth column is the number that matters most for bankroll planning. Martingale busts at 9 losses on this bankroll. Fibonacci busts at 14. That gap is the difference between a 67% chance of a session-ending streak in 1000 bets and a 9% chance. Neither progression changes the expected value. Both lose 1% of total amount wagered on average. What the progression changes is the probability distribution: how likely you are to encounter a streak long enough to wipe the bankroll, and how the recovery path looks once you are in one.
Configuring Payout, Win Chance, and Progression
Configuring a dice bot session means setting five values before the first bet: base bet, win chance, progression type, multiplier on loss, and stop conditions. Every parameter interacts with every other one. A Martingale at 49.5% win chance behaves differently from a Martingale at 10%, because the losing streaks are longer and the recovery wins are rarer at the lower chance. The configuration is not a set of independent knobs. It is a system where each value shifts the risk profile of the whole session.
Win Chance and Variance
Your win chance determines the payout multiplier and the variance of the session. Stake’s default is 49.5% win chance (roll under 50.50), which pays 1.98x and behaves like a near-coin-flip with a 1% edge. Lowering the win chance raises the multiplier but lengthens the losing streaks you will encounter:
| Win Chance | Payout Multiplier | Avg. Loss Streak (per 100 bets) | Risk Profile |
|---|---|---|---|
| 98% | 1.01x | ~1.5 losses | Very low variance, slow grind |
| 49.5% | 1.98x | ~2 losses | Low to medium, default setting |
| 25% | 3.96x | ~4 losses | Medium, longer streaks |
| 10% | 9.9x | ~9 losses | High variance, deep streaks |
| 2% | 49.5x | ~50 losses | Very high, bankroll must absorb long dry runs |
| 0.01% | 9900x | ~6900 losses | Extreme, payout ceiling |
A Martingale at 10% win chance can run 20 or more consecutive losses before a recovery win. Each loss doubles the bet, so 20 losses at 0.10 EUR base means the 21st bet would be 52,428.80 EUR. No realistic bankroll survives that. The same Martingale at 49.5% busts at loss 9 on a 100 EUR bankroll. The progression is identical. The win chance is what changes the depth of the streaks you will encounter, and the depth is what determines whether your bankroll can absorb them.
Base Bet and Bankroll Ratio
Your base bet should be sized relative to your bankroll, not picked arbitrarily. A common starting point is 0.10% to 0.50% of your total balance. On a 100 EUR bankroll, that means 0.10 EUR to 0.50 EUR as the base. The ratio matters because it determines how many consecutive losses your progression can absorb before the next bet exceeds the remaining balance.
On a 100 EUR bankroll with 0.10 EUR base (0.10% of bankroll), Martingale survives 9 losses. With 0.50 EUR base (0.50% of bankroll), Martingale survives only 7 losses, because each doubling step is five times larger. With 1.00 EUR base (1% of bankroll), it survives only 6. The base bet is the lever that controls how deep your progression can go before busting. Smaller base bets mean more survivable sessions but slower recovery and lower per-session profit potential.
Walking Through a Real Session Configuration
Here is a concrete session configuration for a 100 EUR bankroll using Martingale at 49.5% win chance on Stake dice. Every value below is a parameter the bot should expose as a configurable field, not a hidden internal constant.
| Parameter | Value | Why This Value |
|---|---|---|
| Base bet | 0.10 EUR | 0.10% of bankroll, allows 9-loss Martingale survival |
| Win chance | 49.5% | Default Stake setting, ~1.98x payout, lowest variance |
| Progression | Martingale (2x on loss) | Doubles on loss, resets on win, recovers in one bet |
| Max consecutive losses | 8 | Stops one step before bust point (loss 9 costs 25.60 EUR) |
| Max wager per bet | 12.80 EUR | Caps the 8th-loss bet, prevents catastrophic single roll |
| Balance stop-loss | 80 EUR | Halts if balance drops below 80, limiting session drawdown to 20% |
| Max bets per session | 500 | Prevents indefinite running if win target is never hit |
| Session timer | 60 minutes | Hard stop regardless of outcome |
| Profit target | 5 EUR | Stops when session profit reaches 5 EUR, then halts |
The max consecutive losses of 8 is the critical guard. Without it, the bot continues past loss 8, places the 9th bet at 25.60 EUR, and if that loses, places the 10th at 51.20 EUR, which the remaining balance cannot cover. The session ends not because a rule fired but because the balance hit zero. With the cap at 8, the bot stops after the 8th loss having drawn down 12.70 EUR, well within the 100 EUR bankroll, and the session is preserved for another attempt.
The max wager per bet of 12.80 EUR is the second critical guard. It caps the 8th-loss bet at 12.80 EUR (the Martingale value for step 8). Even if the consecutive-loss cap somehow fails, no single bet can exceed 12.80 EUR. Without this, a deep Martingale run can place one catastrophic bet that erases the session in a single roll. For the full breakdown of how layered stop-loss, max wager, and session guards work together in a dice bot, see our Stake dice bot feature guide.
Switching Win Chance Mid-Session
Some bots allow switching between win chances or progression profiles mid-session. This is where configuration depth matters. A session that starts at 49.5% with a Martingale and switches to 10% win chance mid-run is changing the variance profile while the progression is already deep. The losing streaks at 10% are longer, so the Martingale that was safe at 49.5% can wipe the bankroll in three or four losses at the new setting. If the bot supports mid-session switching, the stop-loss and max wager guards must be recalculated for the new win chance before the switch, not after. A bot that lets you switch without recalculating the guards is letting you change the risk profile without adjusting the safety net.
Bankroll Protection in Dice Sessions
Bankroll protection goes beyond a single stop-loss number. It is a set of layered limits that prevent the bot from ever touching more money than you intend to risk in a session. The layers work together. A stop-loss without a max wager can still place one catastrophic bet on a deep Martingale run. A max wager without a drawdown limit can bleed slowly for hours. Real protection means every guard is active at the same time, and that you configured them before the session started, not after you noticed the balance dropping.
The Layers That Matter
- Max wager per bet: the bot will never place a single bet above this amount, regardless of what the progression says. Without this, a Martingale on a deep losing streak can place one bet that erases the session.
- Balance stop-loss: halt when your balance drops below a defined threshold. This is the floor that prevents a bad streak from eating past your allocated bankroll.
- Drawdown shutdown: stop after losing a defined percentage of the session bankroll, not just a fixed number. A fixed stop-loss and a percentage drawdown limit are not the same thing, and you need both.
- Max bets per session: cap the total number of bets regardless of outcome. Without this, a bot that never hits its win target or stop-loss will run indefinitely.
- Max consecutive losses: halt after a set number of losses in a row. At 49.5% win chance, a 10-loss streak is uncommon but not impossible, and a Martingale that deep can wipe a bankroll if no cap is in place.
- Session timer: stop after a configured duration, regardless of results. A hard time limit prevents indefinite running.
- Profit target: stop when you have won a configured amount, then halt. This enforces discipline the way a human would not at 2 AM.
Why the 3 AM Discipline Matters
The single most useful thing about automation is that it will not argue with a stop-loss the way a person does at 3 AM. A human running Martingale manually after 8 losses has only lost 12.70 EUR on this bankroll, which feels manageable, and the temptation to keep going for one more bet is strong. The bot does not negotiate. It hits the configured loss count and stops. But that only works if the stop-loss is set before the first bet, while you are calm and away from the game.
Set the limits while you are calm and treat the numbers as final. The stop-loss should be set based on the progression’s bust point, not on a feeling. Calculate the consecutive-loss depth at which the next bet exceeds your remaining balance, then set the stop-loss one or two steps before that point. For Martingale on a 100 EUR bankroll with 0.10 EUR base, the bust point is loss 9. The stop-loss goes at loss 8. For Fibonacci on the same bankroll, the bust point is loss 14. The stop-loss goes at loss 12 or 13. The bot enforces that number mechanically. For the full survival math at each loss depth, see our Martingale vs Fibonacci comparison.
Session Bankroll Versus Total Balance
A bankroll protection layer that matters more than most people realize is session bankroll allocation. If your Stake account has 500 EUR and you want to risk 100 EUR on a dice session, the bot should operate on 100 EUR, not 500 EUR. Without session bankroll allocation, the bot sees the full balance and the stop-loss is set against the full 500 EUR, meaning it can draw down 400 EUR before the stop-loss fires. With allocation, the bot operates on 100 EUR and the stop-loss fires at the threshold you set against that 100 EUR, not the full account balance. Never leave a bot running with access to more balance than you are prepared to lose in that session.
What Happens When the VPS Reboots
If you run bots overnight on a VPS, crash recovery is part of bankroll protection. When the VPS reboots, the browser crashes, or the network drops mid-session, a bot without crash recovery either starts over from scratch or runs into a corrupted state. A bot with crash recovery detects the interrupted session, restores the last known configuration, and resumes from the correct position in the progression. This is not about recovering losses. It is about not losing your place in a Martingale sequence that was three losses deep into recovery, and not double-placing a bet after a reconnect. Without crash recovery, an interrupted Martingale at loss 6 might restart at the base bet instead of the 6.40 EUR the progression calls for, breaking the recovery math and leaving you stuck between steps with no clear path back.
Browse the StakeProSoft catalogue to compare dice bot builds by the parameters that matter: progression profiles, stop-loss levels, session guards, per-bet wager limits, and 2FA handling. Nothing is a black box. Pre-sales questions are free and a lot cheaper than a licence you cannot use, so reach out before you buy and we will answer them directly. Email [email protected] with your setup details and we will tell you whether the dice bot fits your configuration before you spend anything.
18+ only. Gambling involves financial risk. Bots automate session execution and do not alter the house edge or guarantee profit. Set your limits before the first session and never stake money you cannot afford to lose. Play responsibly.
Frequently Asked Questions
What parameters does a dice bot configure on Stake?
A dice bot configures five core values: base bet, win chance (0.01% to 98%), progression type (flat, Martingale, Fibonacci, or custom), multiplier on loss, and stop conditions (max consecutive losses, balance stop-loss, max wager, session timer, profit target). The payout multiplier is computed as 99 divided by your win chance, so 49.5% pays 1.98x and 2% pays 49.5x. The 1% house edge lives in that 99, not in the roll.
Can a dice bot predict the next roll on Stake?
No. Each dice roll is determined by HMAC-SHA256 using a server seed that is locked before the session starts, your client seed, an incrementing nonce, and a cursor fixed at 0. The server seed is shown only as a SHA-256 hash commitment. No software can reverse, pre-read, or influence the outcome. A bot places the bet and reads the result exactly as you would manually. Any seller claiming their bot predicts RNG is not being truthful.
How does the payout formula work for Stake dice?
The payout multiplier is 99 divided by your win chance. At 49.5% win chance, the multiplier is 1.98x. At 10%, it is 9.9x. At 2%, it is 49.5x. The maximum is 9900x at 0.01% win chance. The expected value per dollar staked is (win chance x multiplier) minus 1, which equals negative 0.01 at every setting. The 1% house edge is constant regardless of which win chance or progression you choose.
Which progression survives longest before busting?
On a 100 EUR bankroll with 0.10 EUR base bet at 49.5% win chance, Fibonacci survives 14 consecutive losses before busting at loss 15. Martingale survives 9 losses before busting at loss 10. In a 1000-bet session, Martingale has roughly a 67% chance of hitting a 10-loss streak, while Fibonacci has roughly a 9% chance of hitting a 15-loss streak. Fibonacci survives longer but needs multiple consecutive wins to recover from a deep streak.
What bankroll protection should I set before a dice bot session?
Configure these layers before the first bet: max wager per bet (caps any single bet regardless of progression), balance stop-loss (halts when balance drops below a threshold), max consecutive losses (stops before the bust point), max bets per session (prevents indefinite running), session timer (hard time limit), and profit target (stops when a configured profit is reached). Set them while calm and treat the numbers as final. The bot will not argue with a stop-loss the way a person does at 3 AM.