Why 100 Quantum Coin Flips Never Give Exactly 50/50
What you see in the Lab
In the quantum coin flip experiment you spin a coin into a 50/50 mix and look at it. Press Try it 100 times and you might see Heads 47 of 100, then 54 of 100. That can feel wrong, because the odds are exactly one half. It is not wrong. It is how random results behave.
The odds are exact, the results are not
The state of the qubit after a Spin the coin step gives Heads and Tails a probability of exactly one half each. A probability describes the long run, not any short run. Each single look is independent of the last one, so there is no force pulling the count back to 50.
How much wobble is normal?
For n independent looks with a one half chance, the number of Heads follows a binomial distribution. Its spread, the standard deviation, is the square root of n times one half times one half. These figures come straight from that formula:
| Number of looks | Typical spread | Chance of exactly half | Chance of landing in the usual range |
|---|---|---|---|
| 100 | about 5 Heads | about 8.0% (exactly 50) | about 96.5% for 40 to 60 Heads |
| 1,000 | about 16 Heads | about 2.5% (exactly 500) | about 96% for 468 to 532 Heads |
So 100 looks give a count between 40 and 60 about 96.5% of the time, and a count between 45 and 55 only about 73% of the time. About one run in 28 falls outside 40 to 60 even with a perfect coin.
The surprising part: more looks, less exact
With more looks the chance of hitting exactly half goes down, from about 8% at 100 looks to about 2.5% at 1,000. What gets tighter is the share: the spread of 16 on 1,000 looks is about 1.6 percentage points, while the spread of 5 on 100 looks is 5 percentage points. That is why the 1,000 button in the Lab lands visibly closer to 50 percent, but almost never exactly on it.
How to tell a biased coin from a fair one
Suppose you get 62 Heads in 100. That is about 2.4 spreads away from 50, which a fair coin produces only a few times in a hundred runs, so it is suspicious but not proof. Press the button again. If the unusual result keeps showing up, something is biased. In the Lab nothing is biased, so the counts keep wandering around 50.
Does a real quantum computer behave the same way?
Yes, the same math applies to the counts from real hardware, with an extra twist: real machines also have noise, so a nominal 50/50 circuit can show a small tilt. The Lab has no noise. It also uses a pseudo random number generator instead of quantum randomness, so it imitates the statistics but is not itself a quantum source.
Try it yourself in Quantum dice, where eight faces share 100 rolls and the bars never come out perfectly even.
Sources and further reading
Checked 2026-10-12. The probabilities and memory figures on this page were calculated directly from the formulas shown. This is education, not investment or security advice. QUANTUM (QNT) is an independent community token on Solana and is not affiliated with Quantinuum (Nasdaq: QNT) or Quant Network (QNT on Ethereum).
Frequently asked questions
Why is my 100 run result not 50 and 50?
Because 100 random looks of a fair coin split exactly 50 and 50 only about 8% of the time. A split between 40 and 60 happens about 96.5% of the time.
Does the Lab give biased results?
No. The simulator uses the exact 50/50 odds of the quantum state, then draws results with a pseudo random number generator, so any wobble you see is normal chance.
Next read
Quantum Random Number Generators: True Randomness You Can CheckHow quantum devices make unpredictable numbers, from phone chips to NIST's CURBy randomness beacon, and why randomness matters for security.Read nextAlso useful: Superposition Explained in Plain English | What Is a Qubit? Superposition and Measurement Explained | Quantum Computing Myths and Misconceptions
Educational content, not financial advice. QUANTUM (QNT) is independent of Quantinuum (Nasdaq: QNT) and Quant Network (QNT on Ethereum). Live chart | Token facts