How Random Is a Spinner Wheel? The Fairness Math

The result is decided before the animation starts

On the spinner wheel, the outcome is determined the instant you click spin, not sometime during the visible deceleration. The tool draws one random angle from the full 360-degree circle using crypto.getRandomValues, then checks which entry's arc contains that angle. Everything you see afterward -- the wheel accelerating, spinning several full rotations, then slowing into its final resting position -- is a cosmetic animation built on top of an already-fixed outcome, purely so the reveal feels satisfying rather than instant and flat.

That distinction matters because it rules out an entire category of concerns people reasonably have about physical spinning objects: there's no friction to be uneven, no click timing to influence, and no way for the visible spin duration to correlate with which entry wins.

Arc-proportion selection, in plain terms

Picture the wheel's full circle as 360 degrees split into segments, one per entry. If you add 6 equal entries, each gets a 60-degree arc; add 10 equal entries and each gets 36 degrees. The random draw picks a landing point somewhere in that 360-degree range, and whichever arc contains the landing point is the winner. Because every degree of the circle is equally likely to be the landing point (the draw is uniform across the full continuous range), every entry's probability of winning is exactly proportional to how much of the circle its arc occupies.

This is the same underlying idea as a fair roulette wheel or a well-built raffle drum -- the geometry does the fairness work, as long as the underlying random draw across the full range is genuinely uniform, which is exactly what crypto.getRandomValues is built to guarantee.

What happens with weighted entries

If you give one entry double the weight of the others, it simply gets double the arc -- twice the degrees of the circle -- which proportionally doubles its odds without requiring any change to how the landing angle itself is drawn. This is a meaningfully different (and arguably more flexible) approach than the rejection-sampling correction most other tools on this site need for their outcome counts, precisely because the wheel's randomness is continuous rather than discrete: there's no leftover uneven range to correct for the way there is when drawing among, say, 20 die faces or 26 letters, since any real number of degrees is a valid landing point.

How this compares to a physical spinner

A physical spinner -- a pointer on a pin, or a literal wheel with a ball -- introduces several small sources of real bias that a digital version doesn't have: an uneven surface can subtly favor certain resting positions, a slightly off-center pin can create a bias toward whichever side has less rotational resistance, and how hard or how consistently a person spins it can, in principle, be learned and exploited over many repeated spins by someone paying close attention. None of those physical variables exist here -- the landing angle is drawn once, mathematically, with no moving parts to introduce a bias.

What happens with a very unbalanced wheel

If you set up a wheel with one entry weighted heavily against several lightly-weighted ones -- say, one entry worth 90% of the wheel and nine tiny entries splitting the remaining 10% -- the fairness math doesn't change at all, even though the visual result might look lopsided. The heavily-weighted entry still only wins in exact proportion to its arc size; it doesn't get any additional boost from being visually dominant on the wheel. This is worth testing yourself: set up a deliberately unbalanced wheel, spin it 50 or 100 times, and tally the results -- the win ratio should track closely to the arc proportions you set, which is a useful way to build direct confidence in the underlying mechanic rather than taking it on faith.

The spinner wheel versus the number wheel

The number wheel, in the numbers category, uses the identical arc-proportion mechanic described above, just auto-populated with a numeric range instead of custom text entries, and with an added optional mode that removes numbers from the wheel once they've been called -- useful for bingo-style live number calling where repeats shouldn't happen. Everything about the underlying fairness math here applies equally to both tools; the difference is purely in what populates the wheel and what happens to entries after they're drawn.

Testing it yourself with a simple experiment

If you want direct evidence rather than taking any of this on faith, set up a wheel with two equal entries labeled A and B, spin it 100 times, and record the tally. A genuinely fair wheel should land close to a 50/50 split, with the kind of small natural variance any fair coin flip would also show over 100 trials -- not a suspicious lean toward one side. Repeat with an intentionally weighted setup (A worth 80%, B worth 20%) and the tally should track that 80/20 split instead. This kind of direct, repeatable test is available to anyone, at no cost, precisely because the tool has nothing to hide about how it works.

Frequently asked questions