FRM Exam Part I · Common Univariate Random Variables

Central Limit Theorem and Standard Error for FRM Part I

Updated 11 October 2026 · Fact-checked

The central limit theorem says that the mean of n independent, identically distributed observations with finite variance is approximately normal when n is large, whatever the original distribution. The sample mean has mean μ and standard error σ ÷ √n. You standardize it with z = (X̄ − μ) ÷ (σ ÷ √n).

Understand Central Limit Theorem

Start with one problem. You draw a sample from a population and compute its average. Draw another sample and you get a different average. The sample mean is itself a random variable, with its own distribution. The central limit theorem (CLT) tells you the shape of that distribution.

The result: take n independent, identically distributed (i.i.d.) observations from a distribution with mean μ and finite variance σ². As n grows, the distribution of the sample mean X̄ approaches a normal distribution with mean μ and variance σ² ÷ n. This holds even if the underlying data are skewed, discrete or otherwise non-normal.

Why does the spread shrink? Averaging cancels out some noise. High values and low values offset each other. The variance of a sum of n independent variables is n σ², so the variance of the average is n σ² ÷ n² = σ² ÷ n. The standard deviation of the sample mean is called the standard error: σ ÷ √n. To halve the standard error you need four times the sample size.

The CLT needs conditions. Observations should be independent (or only weakly dependent), drawn from the same distribution, and the variance must be finite. Fat-tailed data with infinite variance break it. Strong autocorrelation also breaks the simple σ ÷ √n formula. The word 'large' is a rule of thumb: n of about 30 is often enough for mild skew, but heavy skew or fat tails need more. Do not treat 30 as a guarantee.

In practice, you rarely know σ. You replace it with the sample standard deviation s, giving the estimated standard error s ÷ √n. For small samples from a normal population you then use the t-distribution. This is the base for confidence intervals and hypothesis tests on means. Note the CLT is about the average, not about the individual observations: single returns stay as non-normal as before.

Key formulas to remember

CLT distribution of the sample mean: X̄ ≈ N(μ, σ² ÷ n); Holds for large n with i.i.d. observations and finite variance. It is an approximation.
Standard error of the sample mean: SE = σ ÷ √n; Use when the population σ is known.
Estimated standard error: SE = s ÷ √n; Use when σ is unknown and replaced by the sample standard deviation s.
Standardized sample mean: z = (X̄ − μ) ÷ (σ ÷ √n); Look up z in the standard normal table.
Distribution of a sum: ΣXᵢ ≈ N(nμ, nσ²); The sum has standard deviation σ√n, not σ ÷ √n.
Confidence interval for the mean: X̄ ± z × SE; z = 1.645 for 90%, 1.96 for 95%, 2.576 for 99% (two-sided).

How to solve Central Limit Theorem questions

Use this routine for any CLT or standard error question.

1Decide whether the question is about one observation, a sample mean or a sum. This choice fixes the formula.
2Write down μ, σ (or s) and n. Check that σ is a standard deviation, not a variance.
3Check the conditions: independent, identically distributed, finite variance, n large enough. If the population is normal, the result holds for any n.
4Compute the standard error: σ ÷ √n for the mean, or σ√n for the sum.
5Standardize: z = (value − mean) ÷ standard error. Use nμ as the mean for a sum.
6Read the probability from the normal table. For 'greater than', use 1 − N(z).
7If σ is unknown and n is small, use the t-distribution with n − 1 degrees of freedom.
8Sanity check: the answer for a mean should be tighter than the answer for a single observation.

Quickest way: Standard error shortcut

When to use it: Use when the question gives σ and n and asks for a probability or an interval on the sample mean.

Compute σ ÷ √n first. Do it before anything else.
Convert the distance from μ into standard errors: z = distance ÷ SE.
Memorize z of 1, 1.645, 1.96 and 2.576 with probabilities 84.13%, 95%, 97.5% and 99.5% (one-tailed).
To change the sample size by a factor k, the standard error changes by a factor 1 ÷ √k.
Eliminate options that use σ² or σ ÷ n in place of σ ÷ √n.

Common mistakes in Central Limit Theorem

Dividing σ by n instead of √n.
Students confuse variance (σ² ÷ n) with standard deviation.
Fix: Variance of the mean is σ² ÷ n. Take the square root to get SE = σ ÷ √n.
Using the variance as σ in the SE formula.
The question gives σ² and students plug it in without noticing.
Fix: Take √σ² first. Check the units given in the question.
Claiming the individual observations become normal.
The CLT is remembered as 'everything becomes normal'.
Fix: Only the distribution of the sample mean (or sum) approaches normal. The underlying data keep their shape.
Treating n = 30 as a guarantee.
The rule of thumb is taught as a law.
Fix: State it as a guide. Strong skewness or fat tails need larger samples, and infinite variance breaks the CLT.
Using σ ÷ √n for a sum of n variables.
Mixing up the sum with the average.
Fix: The sum has mean nμ and standard deviation σ√n. Divide by n only for the average.
Ignoring dependence in the data.
Return series are assumed i.i.d. by default.
Fix: If observations are autocorrelated, σ ÷ √n understates the true uncertainty. Check the independence condition.

Worked examples

Example 1

Daily P&L of a desk has mean ₹0 and standard deviation ₹2,00,000. Assume 64 independent days. What is the standard error of the average daily P&L, and the probability that the 64-day average exceeds ₹50,000?

Show the solution

SE = σ ÷ √n = 2,00,000 ÷ √64 = 2,00,000 ÷ 8 = ₹25,000.
z = (50,000 − 0) ÷ 25,000 = 2.0.
P(X̄ > 50,000) = 1 − N(2.0) = 1 − 0.9772 = 0.0228.

Answer: Standard error ₹25,000; probability about 2.28%.

Example 2

A bond portfolio's annual return has standard deviation 8%. A risk manager wants the standard error of the mean return to be 1% or less, using independent annual observations. What is the minimum n?

Show the solution

Require σ ÷ √n ≤ 1%, so 8 ÷ √n ≤ 1.
√n ≥ 8.
n ≥ 64.

Answer: At least 64 independent observations.

Exam tips

Always ask: is the question about one observation, the mean or the sum? Wrong-option traps are built on this.
Watch for variance versus standard deviation in the data. Take the square root before dividing.
Expect conceptual questions on conditions: i.i.d., finite variance, and large n. Know that a normal population gives a normal mean for any n.
Questions on sample size usually use SE scaling: quadruple n to halve SE.
Use the calculator's square root key and keep four decimals until the final z lookup.

Practice questions from Common Univariate Random Variables

Central Limit Theorem in other exams

The same ground in other exams, if you are preparing for more than one or want another angle on it.

Central Limit Theorem: frequently asked questions

What are the conditions for the central limit theorem to apply?

The observations should be independent and identically distributed, with a finite mean and variance. The sample must also be large enough for the approximation to be good. Heavier skew or fat tails require larger samples.

How do I calculate the standard error of the sample mean?

Divide the population standard deviation by the square root of the sample size: σ ÷ √n. If σ is unknown, use the sample standard deviation s. The result measures how much the sample mean varies from sample to sample.

Does the CLT mean returns are normally distributed?

No. It says the average of many independent observations is approximately normal. Individual returns can still be skewed and fat-tailed.

What is the difference between standard deviation and standard error?

Standard deviation measures the spread of individual observations. Standard error measures the spread of a sample statistic, such as the mean, across repeated samples. For the mean, SE is the standard deviation divided by √n.