Weighted Random Picking, Explained

What weighting actually does

By default every entry on a Wheel Cue wheel gets one ticket. Ten entries, ten tickets, a 1 in 10 chance each.

Weighting hands out more tickets. Turn on Weighted choices in the wheel options and add a multiplier:

Pizza *3
Sushi Tacos *2 Salad

Pizza now holds three tickets, Tacos two, Sushi and Salad one each — seven tickets in total. The slices on the wheel are drawn from the same numbers, so a bigger slice really is a bigger chance. Nothing is hidden in the animation.

Working out the odds

The arithmetic is a single division. An entry's probability is its weight divided by the total weight.

With the wheel above, the total is 3 + 1 + 2 + 1 = 7:

Two things people get wrong:

*3 does not mean 3× more likely than everything combined. Pizza is three times as likely as Salad, not three times as likely as the rest of the wheel.

Adding an entry changes everyone's odds. Put one more unweighted item on that wheel and the total becomes 8, so Pizza drops from 42.9% to 37.5% without you touching it. Weights are shares of a whole, not fixed percentages.

You never have to do this by hand — the result window reports the exact probability of whatever won.

Hitting a target percentage

To get specific percentages, pick a total that divides cleanly and work backwards.

For a 50 / 30 / 20 split, use a total of 10:

Option A *5
Option B *3 Option C *2

For 70 / 30, use a total of 10: A *7, B *3. For thirds, do not use weights at all — three equal entries already give 33.3% each.

Weights are whole numbers from 1 to 999, so you can go as fine as tenths of a percent (a total of 1000) if you really need to. In practice, anything past a total of 20 is more precision than the decision deserves.

When weighting is right, and when it isn't

Good uses:

• Extra entries in a giveaway. Someone earned five entries: @handle *5 is exactly right, and it is far better than pasting the handle five times because the result window then reports their true probability.
• Raffle tickets. Twelve tickets, weight twelve.
• Honest nudging. Gym *3, Sofa *1 on a decision wheel. Writing the number down often clarifies more than the spin does.
• Unequal chores. Making the horrible job come up less often than wiping the counters is a fair model of the work involved.

Poor uses:

• Simulating skill or merit. A wheel weighted by how much you like someone is not a fair process wearing a hat; it is a decision you made, displayed as a spin.
• Anything you would not publish. If you would not show the audience the weights, do not use them. Undisclosed weighting in a giveaway is the thing people are worried about when they ask whether a wheel is rigged.
• Making a decision come out the way you want. If you find yourself pushing weights until the "right" answer becomes near-certain, you have already decided. Save the spin.

How Wheel Cue picks the winner

The weights become a cumulative table. Pizza *3, Sushi, Tacos *2, Salad becomes the boundaries 3, 4, 6, 7.

One unbiased draw produces a ticket number from 0 to 6, and the code walks the table to find which entry owns it. Tickets 0–2 are Pizza, ticket 3 is Sushi, tickets 4–5 are Tacos, ticket 6 is Salad.

The draw itself uses crypto.getRandomValues() with rejection sampling, exactly as for an unweighted wheel — equal weights are not a special case in the code, they are just a table where every entry owns one ticket.

The full method →