Test the WheelToss draw yourself
Run the draw thousands of times in your own browser and look at the spread — this page calls the same selection code every WheelToss wheel calls.
This is a demonstration you run, not an independent audit. The page calls the same selection code the wheels call, in your browser, and shows you the distribution it produces. Nobody outside WheelToss has certified anything here, and a page shipped by us cannot be evidence about us on its own — its value is that you can run it as many times as you like, and read the code it calls.
For a licensed lottery, a gambling product or clinical randomisation, you need a certified system with an independent auditor behind it. WheelToss is not one and does not claim to be.
Loading the test…
What this test does
Pick a number of slices and a number of draws, and the page runs that many selections through the site’s own draw function, counting how often each slice comes up. It then shows the observed count against the expected count, so a bias would appear as a bar that consistently overshoots or falls short.
The work is split across animation frames rather than run in one blocking loop, so the page stays responsive and the progress bar means something. One hundred thousand draws typically finish in a fraction of a second, phone included.
How to read the result
With equal chances, every slice should land near the same count, and the vertical marker on each bar shows exactly where “the same” is. Small differences are expected and are the whole point of a random process — with 10,000 draws across six slices, a slice landing 1,600 or 1,730 times is entirely ordinary.
What would be a red flag is a pattern: the first slice always high, the last always low, or one slice never reaching its marker across repeated runs. That is the signature of the modulo bias that rejection sampling exists to remove, and it is what this page is here to let you look for.
Turn on ascending weights and the expectation deliberately stops being flat. Slice 3 should land about three times as often as slice 1. A weighted test that came out flat would mean the weights were being ignored.
What it cannot tell you
Statistical testing can reveal obvious bias, but it cannot prove randomness. A generator could pass every test on this page and still be predictable to somebody who knew its internal state. What this page can honestly show is that the selection code has no visible skew, and that the wheel’s animation has nothing to do with the outcome.
It is also our own code testing our own code. That is worth something — you can run it yourself, as many times as you like, on any list size, and you can read the function it calls — but it is not the same thing as an independent review, and we are not going to describe it as one. Nobody outside WheelToss has examined or certified this.
For anything regulated — a licensed lottery, a gambling product, clinical randomisation — you need a certified system with an independent auditor behind it. WheelToss is not one. It is a good tool for classrooms, giveaways, games and everyday choices.
The exact algorithms are written out on the methodology page, and the reasoning behind them is in the guide to how random selection works. See the methodology and the guide to how random selection works.