Stake Dice Strategy: Win Chance, Payout and Risk

On Stake Dice you choose a win chance between 0.01% and 98%, and the game sets the payout. That one slider runs from 1.0102x to 9,900x, and at every position it costs the same 1% of each bet. Here is how the numbers work, and what they do to a session.

How a roll is made

Stake converts your seed pair and nonce into a random number f between 0 and 1, then computes the roll as (f × 10,001) / 100: one of 10,001 results from 0.00 to 100.00. You can recompute any past roll with the script in our provably fair guide.

The payout for a win chance c, in percent, is:

multiplier = 99 / c

So 49.5% pays 2.00x, 90% pays 1.10x and 1% pays 99x. Roll over and roll under are the same bet at the same chance: over 50.50 and under 49.50 each cover 4,950 of the 10,001 results and both pay 2.00x.

Payout, recovery and streaks by win chance

The last two columns give the longest losing streak you should expect in 10,000 rolls, measured over 4,000 simulated sessions per setting (median and 95th percentile).

Why 49.5% is not special

49.5% is popular because it pays a round 2x and suits Martingale. Mathematically it is neither better nor worse than any other setting: the expected loss is 1% of the amount wagered at every point of the slider. What moves is the spread of results. We ran 100,000 sessions of 10,000 flat bets of 1 unit:

At 90% you win nine rolls out of ten and still finish behind in 999 sessions out of 1,000. At 10% most rolls lose, yet more than a third of sessions end ahead, and the bad ones go almost four times deeper than at 90%. Same cost, different ride.

What Martingale actually does

Martingale at 49.5% doubles the bet after each loss and goes back to the base bet after a win. We simulated it 200,000 times: a 1,000-unit bankroll, a 1-unit base bet, stopping at +100 units or when the next bet no longer fits in the balance.

• 87.5% of sessions reach +100.
• 12.5% hit a losing streak the bankroll cannot cover and stop with an average loss of 783 units.
• Averaged over every session, the result is about −10 units for roughly 950 units wagered: about 1% of the action, the house edge again.

That is the whole trap. The system wins seven sessions out of eight and still loses money, because one losing session costs as much as eight winning ones earn. The Stake strategy guide covers the other progressions and the bankroll each one needs.

What an hour of Dice costs

Since the edge is 1% of every bet, the expected cost of a session is simple arithmetic:

expected loss = bet size × bets per hour × 1%

Automation moves you to the right of this table. At bot speed the same bet size costs many times more per hour, and results converge on the expected loss sooner. If you automate, cut the bet size to keep the hourly cost where it was when you played by hand.

Automating Dice

SSPilot's Dice mode takes a win chance from 0.01% to 98%, over or under, a bet rule after wins and losses, and stops on profit or loss. Pick the win chance from the streak you can sit through, size the bet with the hourly-cost table, and set the stop loss first.

How these numbers were produced: payouts follow Stake's published formula. Streak lengths come from 4,000 simulated sessions of 10,000 rolls per setting, session results from 100,000 simulated sessions and the Martingale figures from 200,000, all with a fixed random seed.