Coin flip vs wheel probability: the maths behind every spin

What are the odds on a single spin?

On a wheel with n equal slices, each slice has a probability of 1/n of winning any given spin. The odds of a set of slices are the sum of their chances, so the probability of landing on one of 3 slices out of 12 is 3/12, or 25%.

Each spin is independent: the wheel has no memory, so the chance on the next spin is the same whatever happened before. How SpinDecide makes sure each slice really gets its share is explained in is a spin wheel really random?

Is a two-slice wheel the same as a coin flip?

Mathematically, yes. Two equal slices give the same 50/50 split as an ideal coin, and a sequence of spins behaves exactly like a sequence of flips: same streaks, same long-run averages. The yes or no wheel and the coin flip are interchangeable for a two-way decision.

The practical differences are about convenience. A coin needs you to agree which side means what, while a wheel has the answers written on it. A wheel can also grow a third slice for "maybe", which a coin cannot do. The coin tool, on the other hand, keeps a running count of heads, tails and your longest streak, and can flip up to ten coins at once.

How do weighted slices work?

With weights, each option's probability is its weight divided by the total of all weights:

P(option) = weight of option ÷ sum of all weights

Worked example. A prize wheel has "Sticker" with weight 6, "Pen" with weight 3 and "Mystery prize" with weight 1. The total is 10, so the chances are 60%, 30% and 10%. On SpinDecide the slices are drawn at those same proportions, so the Sticker slice fills 60% of the circle and you can see the odds at a glance.

A useful consequence: doubling every weight changes nothing, because only the ratios matter. Weights of 2, 1, 1 and 4, 2, 2 give the same wheel.

What are the chances of the same result twice in a row?

There are two different questions here, and mixing them up causes a lot of "this wheel is broken" complaints.

• A particular slice twice in a row, decided before you spin: (1/n) × (1/n). On a ten-slice wheel that is 1 in 100.
• Whatever came up last, coming up again: just 1/n, because the first result can be anything. On a ten-slice wheel that is 1 in 10.

For a weighted wheel, the chance of an immediate repeat is the sum of each option's probability squared. On the 60/30/10 prize wheel above, that is 0.36 + 0.09 + 0.01 = 46%.

A related surprise is how quickly any repeat shows up. Spin a ten-slice wheel just four times and the chance that at least two spins match is 1 − (10 × 9 × 8 × 7) ÷ 10,000, which is about 50%. It is the same maths as the famous birthday problem.

How long can a streak get?

Longer than most people expect. In 100 flips of a fair coin, the chance of seeing at least one run of six identical results is about 81%, and a run of seven or more happens a little over half the time. Even in just 20 flips there is roughly a 46% chance of five in a row. A long run on a two-slice wheel is normal behaviour, not a fault.

Does best of three change the odds?

For a fair coin or an equal two-slice wheel, no: each side still has a 50% chance of winning the series. What best of three changes is the feeling. A single flip can seem arbitrary, while a series gives the loser a sense that they had a proper chance. It takes two flips half the time and three flips the other half, so 2.5 on average.

With a weighted wheel, best of three amplifies the favourite's advantage. If one side wins each spin with probability p, its chance of winning best of three is p² × (3 − 2p). At p = 0.6, a single spin gives 60%, but best of three gives 64.8%. Longer series push it further, which is why tournaments use them to let the stronger side show itself.

How many spins until everyone is picked?

This is where wheels and coins part ways in everyday use. Suppose a teacher has 25 names on a wheel and wants every student to be picked. If names stay on the wheel after winning, the expected number of spins before every name has come up at least once is n × H(n), where H(n) is the harmonic number 1 + 1/2 + 1/3 + … + 1/n. This is known as the coupon collector's problem.

The other side of the same coin: with 25 names and 25 spins with replacement, each student has a (24/25)²⁵ chance, roughly 36%, of never being picked. On average about nine students would be left out. Switching on Remove the winner after each spin makes it exactly n spins, because each name leaves the wheel once it wins. The classroom name picker comes set up that way.

Are real coins actually fair?

Nearly, but not perfectly, and the reason is physics rather than the coin's weight. In 2007, Persi Diaconis, Susan Holmes and Richard Montgomery modelled how human hands flip coins and predicted that a flipped coin lands on the same side it started on about 51% of the time, because real flips include a slight wobble.

In 2023, František Bartoš and a large team of collaborators tested this with 350,757 recorded flips. Coins landed on their starting side about 50.8% of the time, closely matching the prediction, and they found no meaningful difference between heads and tails themselves. The bias varied between flippers and seemed to shrink with practice.

In practice, 50.8% only matters if someone can see which side is up before the flip and exploit it. A digital coin or wheel has no starting side, no wobble and no hand, so its 50/50 is exact up to the quality of the random generator.

Coin or wheel: which should you use?