Skip to content

CFA Level I Exam · Statistical Distributions for Financial Asset Prices and Returns

Monte Carlo Simulation: Steps, Uses and Limitations for CFA Level 1

Updated 7 October 2026 · Fact-checked

Monte Carlo simulation repeatedly draws random values for risk factors from assumed probability distributions, runs each draw through a model, and collects the results into a distribution of outcomes. You then read off the mean, standard deviation or percentiles. Its output is only as good as its assumptions.

Understand Monte Carlo Simulation

Many finance problems have no neat formula. A path-dependent option, a pension fund's funding ratio, or a portfolio's loss over a year may depend on several uncertain inputs at once. Monte Carlo simulation handles this by brute force. You describe how each uncertain input behaves, let a computer draw random values, and compute the result many times.

Each run is called a trial. In one trial, the computer draws a value for every risk factor, such as a stock return or an interest rate, and plugs the values into your model to get one outcome. After thousands of trials you have thousands of outcomes. Their histogram approximates the distribution of the result, and you can estimate its mean, standard deviation, and tail percentiles.

The draws come from a pseudo-random number generator that produces numbers between 0 and 1. These are converted into draws from the chosen distribution. For a normal variable, for example, a random number is mapped to a standard normal value and then scaled by the assumed mean and standard deviation. If risk factors are correlated, the simulation must reflect that correlation, or the results will understate or overstate risk.

Compare this with historical simulation. Historical simulation samples from actual past observed changes, so it assumes the future resembles the past and needs no distribution assumption. Monte Carlo uses a distribution you specify, so it can produce scenarios never seen in the data. The cost is that you must choose the distribution and its parameters.

Uses include valuing complex securities such as options with no closed-form price, estimating value at risk, testing a portfolio or retirement plan under many scenarios, and assessing how sensitive a result is to its inputs. Key limitations: it is a statistical, not analytic, method, so it gives estimates and no cause-and-effect insight. It is complex and computationally heavy. It depends on the model and input assumptions, so it suffers from garbage in, garbage out. The standard error of the estimate shrinks as trials increase, but more trials never fix a bad model.

Key formulas to remember

Standard error of a simulated mean
Standard error = s ÷ √N
s is the standard deviation of the simulated outcomes and N is the number of trials. Quadrupling the trials halves the standard error.
Converting a standard normal draw
X = μ + σ × Z
Z is a random draw from the standard normal distribution. X is a draw from a normal variable with mean μ and standard deviation σ.
Lognormal price draw (one step)
S(t+Δt) = S(t) × exp(r), where r is the simulated continuously compounded return
Simulating the return as normal keeps the price positive.
Estimate from simulation
Estimated value = average of the N simulated outcomes
For a derivative, average the discounted payoffs. Percentile estimates read from the sorted outcomes give measures such as VaR.

How to solve Monte Carlo Simulation questions

Use this sequence for any Monte Carlo question, whether it asks for the steps, a comparison or a limitation.

  1. 1Identify the quantity you want to estimate, such as a price, a loss or a funding ratio.
  2. 2List the risk factors that drive it and decide the distribution and parameters for each, including any correlations.
  3. 3Generate random draws for every risk factor and compute the outcome for each trial using the model.
  4. 4Repeat for a large number of trials, then summarise: average, standard deviation, percentiles.
  5. 5Check precision: the standard error falls with √N, so more trials narrow the estimate only.
  6. 6Match the question to the right idea: if it asks about assumptions or realism, the limit is the model and inputs; if it asks about unseen scenarios, Monte Carlo gains over historical simulation.
  7. 7Eliminate options that claim simulation gives exact answers, proves causes, or needs no assumptions.

Quickest way: Three-check elimination for Monte Carlo MCQs

When to use it: Use this for conceptual three-option questions on steps, uses, or limitations when time is short.

  1. Ask: does the option say the result is exact or free of assumptions? If yes, it is wrong.
  2. Ask: does the option say historical simulation uses an assumed distribution, or Monte Carlo uses only past data? If yes, the roles are swapped, so it is wrong.
  3. Ask: does the option say more trials fix model error? If yes, it is wrong; more trials only cut sampling error.
  4. For numbers, use standard error = s ÷ √N and check the direction: more trials give a smaller value.

Common mistakes in Monte Carlo Simulation

  • Saying Monte Carlo draws from historical data

    It is confused with historical simulation and bootstrap resampling.

    Fix: Monte Carlo draws from a specified distribution. Historical simulation draws from observed past changes.

  • Believing more trials make the model accurate

    Students mix sampling error with model risk.

    Fix: More trials reduce sampling error only. Wrong distributions, parameters or correlations still give wrong answers.

  • Treating the output as a precise answer

    Computer output looks exact.

    Fix: It is a statistical estimate with a standard error, and it provides no analytic cause-and-effect insight.

  • Ignoring correlation among risk factors

    Each draw is thought of as independent.

    Fix: If factors move together, the simulation must draw them with the right correlation, or tail risk is misstated.

  • Forgetting to discount payoffs when valuing a derivative

    Students stop at the average payoff.

    Fix: Discount each payoff, or the average payoff, back to today before reporting a value.

Worked examples

Example 1

An analyst runs a Monte Carlo simulation of a portfolio's one-year return with 400 trials. The standard deviation of the simulated returns is 12%. What is the standard error of the estimated mean return? A) 0.6% B) 3.0% C) 12.0%

Show the solution
  1. Use standard error = s ÷ √N.
  2. √400 = 20.
  3. 12% ÷ 20 = 0.6%.

Answer: A) 0.6%

Example 2

A risk manager says: 'We simulated 100,000 trials, so our model of asset returns must be correct.' Which response is most accurate? A) The statement is correct because a large number of trials removes all error. B) The statement is wrong because many trials reduce sampling error only; the distribution and parameters assumed may still be wrong. C) The statement is wrong because Monte Carlo simulation uses only historical returns.

Show the solution
  1. Separate the two error types: sampling error and model error.
  2. Many trials shrink sampling error through the √N effect, so A overstates.
  3. C is wrong because Monte Carlo draws from an assumed distribution, not only history.
  4. B correctly says the assumptions can still be wrong.

Answer: B

Exam tips

  • Know the sequence: specify the model and distributions, draw random values, compute outcomes, repeat, then summarise.
  • Contrast Monte Carlo (assumed distribution) with historical simulation (actual past data) every time you see both.
  • Limitations are a frequent question: it is an estimate, it needs assumptions, it is complex, and it gives no analytic insight.
  • For standard error questions, use s ÷ √N and remember numerical options run from smallest to largest.

Practice questions from Statistical Distributions for Financial Asset Prices and Returns

Monte Carlo Simulation in other exams

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

Monte Carlo Simulation: frequently asked questions

What are the steps in a Monte Carlo simulation?

Specify the model and the distributions of the risk factors, then generate random draws for them. Compute the outcome for each trial, repeat many times, and summarise the results with a mean, standard deviation or percentiles.

How is Monte Carlo simulation different from historical simulation?

Monte Carlo draws from a distribution you assume, so it can create scenarios not seen before. Historical simulation draws from observed past changes and needs no distribution assumption, but it assumes the past represents the future.

What are the limitations of Monte Carlo simulation?

It gives statistical estimates, not exact answers or analytic insight. Results depend on the assumed model, distributions, parameters and correlations. It can also be complex and computationally demanding.

Does a larger number of trials improve Monte Carlo results?

It improves precision, because standard error falls in proportion to 1 ÷ √N. It does not correct a poor model or bad inputs.