Equità e caso
Ogni estrazione su ClassCaddy proviene da un'unica implementazione verificata — caso crittografico con campionamento a rigetto e vincitore deciso prima dell'animazione. Non crederci: provalo qui sotto.
How we draw randomness
Browsers ship a cryptographically secure random number generator, crypto.getRandomValues, designed for security-critical use. Every pick on this site — wheels, groups, generators, raffles — flows through one small module that calls it.
Two details matter:
- Rejection sampling: raw random integers have modulo bias when the range doesn't divide evenly; we discard the biased remainder so all options are exactly equal.
- Decide-then-animate: the winner is fixed the moment you press spin; the animation is a truthful retelling, never the decider.
Run your own audit
The test below runs the exact function the tools use, up to 250,000 draws, and computes a chi-square goodness-of-fit statistic against a uniform expectation. A uniform random source lands under the critical value about 95% of the time — so an occasional “investigate” result on an honest wheel is statistics working normally. Run it five times; that's the real test.
FAQ
Is the wheel really random or does it repeat winners?
Every spin calls the browser's cryptographic random source (crypto.getRandomValues) — the same generator used for security tokens. There's no streak logic, no “recently picked” memory, and no weighting unless you set weights yourself.
Why can a name repeat several times in a row?
True randomness produces streaks — if you flip a fair coin 100 times, runs of 5+ heads happen more often than intuition suggests. A picker that never repeats is not random; it's a queue. If you want a queue, enable Remove-winner mode.
What is the chi-square test below?
It draws your chosen number of samples and measures whether the spread across options is within ordinary random variation. At the 0.05 significance level, a fair wheel passes about 95% of the time — failing once in twenty runs is expected behavior, not proof of bias.
What's rejection sampling?
Computers produce random 32-bit integers, but turning those into, say, a choice of 7 leaves a remainder that favors some values (modulo bias). We reject and redraw any number that falls in the remainder range, so every option keeps exactly equal probability.
Was the winner decided before the animation?
Yes — the winner is selected the instant you press SPIN, and the animation is choreographed to land on it. That guarantees the pretty spin can never influence the outcome.