Spunk Odds Reference

Crash guide

Bamboo Crash: the waiting time for a 5.00x round

Bamboo is the label on this address, not a variant with its own rules. The page asks a single practical question about the crash format: if you aim for 5.00x, how many rounds will you usually wait for the first success?

Waiting is a geometric process

When each round is independent and the success chance per round stays fixed, the number of rounds until the first success follows the geometric distribution. In the site model at a 1% edge, a 5.00x target succeeds with probability p = 0.99 / 5 = 19.80%. The chance that the first success comes on round k is (1 - p)k-1 x p, and the chance it has come by round k is 1 - (1 - p)k.

Waiting for the first 5.00x round at a 1% edge (p = 0.198 per round)
Round kP(first 5.00x success exactly at k)P(first 5.00x success by round k)
119.800%19.80%
312.735%48.42%
58.191%66.82%
102.718%88.99%
150.902%96.35%
250.099%99.60%
400.004%99.99%

Mean, median and the long tail

The mean wait is 1/p = 5.05 rounds. The median is smaller: by round 4 the chance of at least one success first passes one half. The gap between mean and median comes from the tail. A few very long waits pull the average up, while most waits are short.

The tail is not tiny. The chance of going 15 rounds with no 5.00x round is 3.65%. The chance of going 25 rounds without one is 0.40%. To be 99% sure of seeing a success you must allow 21 rounds. Each of those rounds costs a stake while you wait.

The memoryless property

The geometric distribution has a property that surprises many people: it has no memory. Suppose you have already missed 8 rounds in a row. The chance that the next round succeeds is still 19.80%, and the expected number of further rounds is still 5.05. Rounds already lost do not count toward the next success, because each crash point is drawn fresh.

This is where the feeling of being "due" goes wrong. The table shows that a success by round 10 is likely in advance. It says nothing about round 11 once rounds 1 to 10 are known misses. Conditional on those misses, the clock starts again from zero.

Comparing waits across targets

A lower target shortens the wait and a higher one stretches it, but the cost per round is the same 1% of the stake in this model. For comparison, a 2.00x target has a mean wait of 2.02 rounds, while a 20.00x target has a mean wait of 20.20 rounds. A longer wait means more stakes are placed before each win, and each win is larger to match, so the averages balance.

Does the wait get shorter after a long dry spell?

No. In a model with independent rounds the expected remaining wait is always 1/p, no matter how many misses came before.

Why is the median below the mean?

The distribution is skewed to the right. Most waits end early, but a few run very long and lift the average.

Is the wait the same thing as the cost?

No. Each round you wait has an expected loss of 1% of the stake in this model, so a long wait is expensive even when the win arrives in the end.