Definition
A random seed initializes a pseudo-random number generator so repeated runs with the same inputs produce the same sampling sequence.
Random seed
A value that controls reproducibility of random sampling.
Key points
- Fixed seeds make Monte Carlo results reproducible.
- Different seeds give slightly different estimates for the same sample size.
- Larger samples reduce the dependence on seed.
Common misconceptions
Myth
Changing the seed changes the true equity.
Reality
It changes the estimate, not the underlying probability.
Myth
A fixed seed is always better.
Reality
It’s better for debugging; random seeds are better for unbiased sampling.