FAIR CHOICES, LOCAL DRAWS

Secure random number generator

Roll, draw or sample. Every value stays within the rules you set.

Your random result

● LOCAL
GENERATED ON YOUR DEVICE
Checking secure randomness…
Custom dice ↓
Options & batch generation

Changing an option replaces the current results. Nothing is saved or sent.

    Entropy & possibilities

    Generated entirely in your browser.
    Nothing is sent to our server.

    Inclusive endpoints, no modulo bias

    The minimum and maximum are both possible results. The supported bounds are signed 32-bit integers: −2,147,483,648 through 2,147,483,647. The range may therefore contain up to 2³² choices. Rejection sampling avoids the bias introduced by taking a random byte modulo an arbitrary range size. The d6 and d20 presets simply select ranges 1–6 and 1–20; enter 1 and any supported maximum for custom dice.

    Unique draws are sampling without replacement

    Unique mode selects from a shrinking virtual pool, so requesting every value in a small range finishes without repeatedly discarding duplicates. If you request more unique results than the range contains, the tool explains the conflict. Sorting changes the presentation and the uncertainty of the entire output: the entropy panel describes the whole ordered batch or sorted set, not a single number.

    Simulation and unpredictability

    Web Crypto is intended for unpredictable choices. It does not provide a user seed or reproducible sequence, so a seeded simulation library may be more suitable when experiments must be repeated exactly. This page is useful for local draws and tests; it is not an audited lottery, proof of fairness or cryptographic protocol by itself.

    Useful to know.

    Is the coin fair?

    Coin mode draws uniformly from two outcomes and displays 0 as Heads and 1 as Tails. Independent flips can repeat; a short run of identical outcomes is possible.

    Why does sorting reduce the entropy shown?

    Multiple draw orders can collapse to the same sorted output. Unique sorted sets have an exact combinatorial count. With repeated values, sorted outcomes are not equally likely, so the displayed estimate is a conservative lower bound.