DiceDecide

How to Calculate the Chance of Exactly k Heads in n Coin Flips

Be the first to rate this page.

The binomial coefficient C(n,k) is the number of distinct ordered sequences containing exactly k successes among n trial positions.

A worked example with numbers

Take eight fair flips and ask for exactly three heads. First choose the three positions: C(8,3) = 8! ÷ (3! × 5!) = 56. Each one fully specified eight-flip sequence has probability (1/2)^8 = 1/256. Multiplying 56 sequences by 1/256 gives 56/256 = 0.21875, or 21.875%. The order H-T-H-T-T-H-T-T counts, but so does every other placement of three H values.

How to set the rule before the result

For an unfair but stable coin with heads probability p, use C(n,k) × p^k × (1-p)^(n-k). Set n to the number of flips actually planned, k to the exact count, and do not replace “exactly” with “at least.” The multiple coin flip tool can display a concrete sample; it does not replace a formula for every possible sample. A probability calculator is useful only after its event and assumptions match the question.

Common mistakes that change the odds or the process

A common mistake is using only (1/2)^k. That gives the probability of heads in k named positions, not exactly k heads anywhere among n positions. Another is treating 3 heads in 8 flips as the same as 3 or more heads; the latter adds probabilities for 4, 5, 6, 7, and 8 heads. Do not round early: 56/256 is exact, while 0.22 is only a display approximation.

Where this method stops being appropriate

This formula needs n independent trials and a constant probability p for each trial. It does not describe draws without replacement, a coin whose mechanism changes, or a game rule that stops as soon as a target count occurs. It also cannot decide whether a short observed sample proves a coin fair. Use it to calculate the chance under explicit assumptions, not to manufacture certainty from a result.

How the random source fits into the rule

NIST’s binomial-distribution reference states the binomial probability function for n independent experiments with success probability p. Its fair-coin special case sets p = 1/2. The implementation source for a browser’s strong random integers is MDN’s Crypto.getRandomValues reference; source quality and the combinatorial count are separate questions.

Read the formula’s three parts

C(n,k) counts placements, p^k gives the probability of the k head positions, and (1-p)^(n-k) gives the probability of the tail positions. For a fair coin p = 1/2, so every full sequence has the same probability 1/2^n. The coefficient is necessary because “exactly k” includes many sequences. In ten flips with four heads, C(10,4) = 210, not one, is the number of accepted placements.

Make a small table before using software

Event in 4 fair flipsFavourable sequencesProbability
Exactly 0 heads11/16 = 6.25%
Exactly 2 heads66/16 = 37.5%
Exactly 4 heads11/16 = 6.25%

The entries total 16 when all five possible head counts are included, matching 2^4 ordered sequences.

Exactly, at least, and at most are different requests

For at least three heads in four flips, add exactly three and exactly four: 4/16 + 1/16 = 5/16 = 31.25%. For at most one head, add exactly zero and exactly one. These additions work because the exact-count events cannot occur together in the same four-flip sample. Write the inequality before calculating so a familiar formula does not answer the wrong question.

Explain the answer with its assumptions

Say “exactly 3 heads in 8 independent fair flips has probability 21.875%” rather than “three heads is 21.875%.” The first statement names the count, trial number, and model. It also makes clear why a different sample size, a loaded coin, or a stopped-early game requires a different calculation.

Check the distribution totals one

For any fixed n, probabilities for exactly 0 through exactly n heads must sum to 1. With n = 3, the probabilities are 1/8, 3/8, 3/8, and 1/8. This is a useful arithmetic check: the coefficients 1, 3, 3, 1 add to 8, which is 2^3. If a table of mutually exclusive exact counts does not total 100%, a coefficient, denominator, or rounding step is wrong.

Use fractions while adding several exact cases, then convert the final answer to a percentage. That practice keeps a calculator display from hiding an error such as adding 0.333 and 0.333 before the full event is defined.

Use a reproducible notation

Write the calculation as C(8,3)(1/2)^3(1/2)^5 = 56/256. A reader can see both the number of placements and the probability of each placement. This notation also makes the generalisation to another coin probability p visible without changing the event description.

For a check, swap heads and tails: exactly three heads in eight fair flips has the same probability as exactly five heads, because every sequence has a mirrored sequence. Symmetry is a useful way to spot an incorrect coefficient near the two sides of the distribution.

Related DiceDecide tools

Enter your values, review the result, then use it with confidence.

Rate this page

Be the first to rate this page.