Stake Limbo Strategy: Targets, Odds and Losing Streaks

Limbo has one setting that matters: the target multiplier. Every target costs the same, and the target only decides how often you win and how long you wait between wins. The numbers below come from Stake's published formula and from simulation.

Where the Limbo result comes from

For each bet, Stake turns your seed pair and nonce into a random number f between 0 and 1 (the mechanism is in our provably fair guide), then computes:

result = max(floor(0.99 / f × 100) / 100, 1.00)

You win when the result is at or above your target. Because f is uniform, the chance of reaching a target t is exactly:

P(win) = 0.99 / t

At 2.00x that is 49.5%; at 100x, 0.99%. The 0.99 is the 1% house edge: a fair 2x bet would win half the time, Limbo pays it 49.5% of the time. Multiply any target by its win chance and you get the same 0.99, the share of each stake that comes back on average.

Win chance and losing streaks by target

The most useful column is the last one: the longest losing streak you should expect in a session of 10,000 bets. We measured it over 4,000 simulated sessions per target and report the median and the 95th percentile.

Streaks grow much faster than the average wait. At 100x you win once every 101 bets on average, yet in a 10,000-bet session a drought of about 500 bets is the normal case, not bad luck. And 10,000 bets is not a long session once a bot places them: at a few bets per second it is an hour or less.

Same cost, different sessions

We simulated 100,000 sessions of 5,000 flat bets of 1 unit at four targets:

Every target loses 1% of what was wagered, 50 units on 5,000. Low targets make that loss steady and close to certain. High targets make it lumpy: more winning sessions, and much deeper losing ones. There is no target where the edge goes away.

Size the bet against the streak, not the win rate

Take the 95th-percentile streak for your target and session length, and check what that many losses in a row costs at your bet size. At 10x, that is 94 bets: with a 10-unit bet, a normal bad stretch takes 940 units; at 1 unit, 94. If that number is more than you would calmly lose, the bet is too big, however often the target seems to hit.

Progressions change this completely. In a 10,000-bet session at 2.00x, a streak of 14 losses or more happens 29% of the time, and 16 or more 8.5% of the time (exact probabilities). A 14-loss streak costs a Martingale 16,383 base bets, and the next bet it asks for is 16,384; after 16 losses the bill is 65,535 base bets. Limbo at 2.00x is mathematically the same bet as Dice at 49.5%, so the Martingale simulation in our Dice guide applies here unchanged, and the Stake strategy guide lists the bankroll every common progression needs.

Predictors and patterns

Each result depends only on the seed pair and the nonce, and the server seed stays hidden behind its hash until you rotate it. Past results carry no information about the next one: 200 bets without a 100x does not make a hit due. Tools that claim to predict Limbo are reading noise, and the ones that ask for your account credentials are a security risk.

Automating Limbo

SSPilot's Limbo mode takes a target from 1.01x to 1,000,000x, a bet rule after wins and losses, and stops on profit, loss or a number of rounds. Choose the target by the streak you can sit through, not by the payout you want, and set the stop loss before the first bet.

How these numbers were produced: win chances are exact from Stake's published formula. Streak lengths come from 4,000 simulated sessions of 10,000 bets per target, the session table from 100,000 simulated sessions of 5,000 bets, all with a fixed random seed. Streak probabilities at 2.00x are computed exactly.