
A spinner wheel looks like physics: you flick it, it slows down, it stops. But an online wheel is a small computer program, and whether it is fair depends entirely on how that program picks the winner. This guide explains how a fair online wheel works, why it still produces streaks and repeats, and how you can check any wheel yourself.
The short answer
A well-built online wheel is more random than a physical one. A cardboard wheel can stick on a slightly heavier side, catch on a worn peg, or be spun the same way by the same practiced hand. A digital wheel that draws its result from a cryptographic random number generator gives every slice exactly its fair share, every time.
The catch is the phrase “well-built”. Some wheels pick a winner badly, and some animations look random while hiding a bias. Here is what to look for.
Two ways to build a wheel
There are two common ways to program a spinner wheel.
1. Simulate the spin, then see where it lands. The program picks a random starting speed, runs a little physics simulation with friction, and reports the slice under the pointer when the wheel stops. This feels natural, but fairness depends on the details. If the random speed is not spread evenly, or friction is modeled so that certain stopping angles are more common, some slices win more often.
2. Decide the winner first, then animate. The program draws the winning slice directly, with the right probability for each slice, and then calculates an animation that lands on it. The animation can be as dramatic as you like, because it cannot change the outcome.
Our wheels use the second method. When you press Spin, the winner is drawn from your browser’s cryptographic random generator before the wheel moves. The spin time, the speed of your flick, and quick mode change only how the result is shown.
Where the randomness comes from
Every modern browser includes a function called crypto.getRandomValues(). It is the same source browsers use to create encryption keys, and it is seeded from unpredictable events inside your device. Its output cannot be predicted, even by someone who has seen previous results.
Many simple websites use a different function, Math.random(), which is fast and fine for games but is not designed to be unpredictable. It is also easy to use incorrectly, which brings us to the most common hidden bias.
Modulo bias: the quiet way wheels go wrong
Suppose a random generator gives you a number from 0 to 255, and you want to pick one of 10 slices. The obvious trick is to take the remainder after dividing by 10. But 256 is not a multiple of 10. The numbers 0 to 249 split evenly into 25 groups of 10, and the six leftover values (250 to 255) map to slices 0 to 5. Those six slices now win slightly more often than the other four.
With bigger random numbers the bias gets tiny, but it never disappears. The fix is called rejection sampling: if the random number lands in the leftover range, throw it away and draw again. Every slice then has exactly the same chance. Our tools use rejection sampling for every pick, from wheels to dice to card draws.
Why fair wheels still repeat themselves
Even a perfectly fair wheel will land on the same entry twice in a row, sometimes three times. That is not a flaw. Each spin is independent; the wheel has no memory of the last result.
Our intuition about repeats is usually wrong. With 10 equal entries and only 5 spins, the chance that at least one entry comes up twice is about 70% (69.8%, to be exact). It is the same math as the famous birthday problem, where a room of just 23 people has a better than even chance (50.7%) that two share a birthday.
| Entries on the wheel | Spins | Chance of at least one repeat |
|---|---|---|
| 10 | 5 | 69.8% |
| 8 | 3 | 34.4% |
| 25 | 20 | almost certain |
So if a wheel of class names picks the same student twice in a short session, that is normal randomness at work. If you want everyone to get a turn, turn on Remove the winner so each name leaves the wheel after it wins.
Streaks feel suspicious, but they are expected
The same goes for streaks. Spin a two-slice yes or no wheel 10 times, and there is roughly a 47% chance of seeing at least four of the same answer in a row. Over 100 spins, a run of six or more in a row happens about 81% of the time.
People tend to expect random results to alternate more than they really do. When asked to write down a “random” sequence of coin flips, most people avoid long runs, which is exactly how researchers can often tell a human-made sequence from a real one.
Weighted wheels are fair when the weights are visible
A weight changes how likely an entry is to win. On a fair weighted wheel, the slice size changes too, so an entry with weight 2 has a slice twice as wide as an entry with weight 1, and wins twice as often. Nothing is hidden: the odds are what you see.
Be wary of any tool that offers a “rigged” or “secret” mode, where a chosen entry wins regardless of what the wheel shows. That is not a random wheel any more. Our wheels have no hidden modes, and the code that picks a winner is the same for every spin.
How to test a wheel yourself
You do not have to take anyone’s word for it. There are two simple tests.
The quick test. Our tools include a Test the fairness yourself box. Press Run 10,000, and the tool draws 10,000 results on your device using the same random function as the real spins. It shows each entry’s share next to its expected share, plus a chi-square p-value. A p-value above about 0.01 means the spread is consistent with a fair draw. Run it a few times: about one run in a hundred will look unusual by pure chance, and that is expected too.
The manual test. For any wheel, spin it many times and tally the results. With 4 equal slices and 400 spins, each slice should land around 100 times. Expect some wobble: anything from about 80 to 120 is normal for a fair wheel. A slice that keeps landing far outside that range over several runs of 400 is worth questioning.
What a fair wheel cannot do
A random wheel makes a fair choice between the options you give it. It cannot make the options fair. If one entry is “Pizza” and another is “Pizza with extra cheese”, the pizza idea has two chances. If your raffle list includes a name twice by mistake, that person has double the odds. Clean the list first: use Remove duplicates, and check the entry count before an important spin.
Checklist for a trustworthy spin
- Use a wheel that decides the winner with cryptographic randomness.
- Make the odds visible: equal slices, or weights that change slice size.
- Clean the list: no duplicates unless they are intended.
- For draws where everyone should win once, remove each winner.
- For public draws, spin live or record the screen, and keep the result history.
Try it
Open the spin the wheel tool, add a few entries, and press Run 10,000 in the fairness box. Then try the number wheel with a large range and watch how often repeats appear in just a handful of spins. Seeing randomness in action is the fastest way to trust it.


