Spunk Odds Reference

Wheel guide

Bamboo Wheel: a seeded simulation of wheel rounds

Bamboo is the word attached to this address, and it does not mark a different wheel design. The page simulates wheel rounds on a computer and explains what such a simulation can and cannot show.

The wheel being simulated

The illustrative set has 16 segments: seven 0x, five 1.5x, two 2x, one 3x and one 1x that simply returns the stake. Its mean return is 0.96875, an expected net of -3.12% per round. That exact number is the benchmark for the simulation.

Illustrative 16-segment wheel used in the Bamboo simulation
MultiplierSegmentsChanceContribution to mean
0.00x743.75%0.0000
1.00x16.25%0.0625
1.50x531.25%0.4688
2.00x212.50%0.2500
3.00x16.25%0.1875

Mean return 0.9688 per unit (EV -3.12%); standard deviation 0.943 per spin.

How the simulation works

For each simulated round, the program draws a whole number from 0 to 15 with Python's pseudo-random generator, takes the segment at that position, and records the multiplier minus one as the net result. The generator starts from seed 4417, so the run is reproducible. Six sessions of 120 rounds are played, one unit per round.

Simulation: Bamboo wheel, 6 sessions of 120 one-unit rounds
SessionRounds wonNet (units)Lowest pointLongest losing run
166-11.00-11.004
265-1.00-10.507
361-6.00-12.005
464-15.50-18.506
562-16.50-16.504
672-1.00-3.005

Seeded pseudo-random simulation (Python random.Random, seed 4417), 6 sessions of 120 one-unit rounds. Average session result -8.50 units. This is a simulation, not a record of real play; a different seed gives different numbers.

Reading the output

"Rounds won" here counts every round that did not lose money, which includes the 1x segment, since it returns the stake with a net of zero. The expected count is about 67.5 per session. The net column is the session total. The theoretical average for 120 rounds is -3.75 units, with a standard deviation of roughly 10.33. Any session in the table that sits within two standard deviations of the average is unremarkable.

The lowest-point column is the most practical one. It shows how far a session dipped below the start before it ended. Two sessions with the same final net can have very different low points, and the low point decides whether a limited bankroll would have survived.

Checking the simulator against the exact value

A simulation is only useful if it is correct. The simplest test is a long run. Using the same simulator and seed for 300,000 rounds, the average net per round is -0.03406. The exact value is -0.03125. They agree to a few thousandths, a gap of about two standard errors for that many rounds. A large gap would point to a bug, such as an off-by-one in the segment index.

Good habits for simulation