Chapter 8

Monte Carlo Simulation

📖 ~50 min read 📈 2 interactive charts ✍️ 2 practice questions 🎯 1 linked game

8.1 Why Simulate?

Some business problems are too complex for neat analytical solutions. When a decision depends on multiple uncertain inputs — each following its own probability distribution — closed-form formulas often do not exist. Monte Carlo simulation offers a powerful alternative: instead of solving for the answer mathematically, you generate thousands of random scenarios and let the distribution of outcomes speak for itself.

A Brief History

The technique takes its name from the famous Monte Carlo Casino in Monaco. During World War II, physicists Stanislaw Ulam and John von Neumann used random sampling to model neutron diffusion in nuclear weapons research. They needed a code name for the classified work and chose "Monte Carlo" as a nod to the games of chance played at the casino. The method has since spread to virtually every quantitative field: finance (pricing options and measuring portfolio risk), engineering (reliability analysis), supply chain management (demand forecasting under uncertainty), healthcare (clinical trial design), and many more.

When to Use Monte Carlo

  • Multiple uncertain inputs: When two or more variables are uncertain and interact to produce an outcome.
  • No closed-form solution: When the mathematical relationship is too complex for algebraic solution.
  • Risk quantification: When you need to understand the full range of possible outcomes, not just the expected value.
  • Decision support: When stakeholders want to see the probability of hitting a target, breaking even, or incurring a loss.
🏪 NorthStar Enterprises

NorthStar is evaluating a new product launch. Revenue depends on three uncertain factors: customer demand, the unit selling price (which varies by channel and negotiation), and the unit production cost (which depends on raw material prices and supplier terms). Fixed costs for the launch are $150,000. Management wants to know: What is the probability that this product will be profitable? What is the expected profit? And what is the worst-case scenario?

No single formula can answer all three questions simultaneously when each input is uncertain. Monte Carlo simulation can.

✓ Check Your Understanding
Monte Carlo simulation is most appropriate when:
The problem has a simple closed-form solution
Only one input variable is uncertain
Multiple uncertain inputs interact and you need the full distribution of outcomes
You only need the expected value of the outcome

8.2 Building a Monte Carlo Simulation

A Monte Carlo simulation follows five systematic steps: (1) define the input distributions, (2) generate random values from each distribution, (3) calculate the output for each set of random inputs, (4) repeat thousands of times, and (5) analyze the resulting output distribution.

Step 1: Define Input Distributions

Each uncertain input must be modeled with an appropriate probability distribution. The choice of distribution depends on what you know about the variable:

  • Normal distribution: Use when the variable clusters symmetrically around a mean. Suitable for demand when you have historical data showing a bell-shaped pattern.
  • Uniform distribution: Use when all values in a range are equally likely. Suitable when you only know the minimum and maximum possible values.
  • Triangular distribution: Use when you can estimate a minimum, most likely, and maximum value. Common for cost estimates and expert judgments.

NorthStar Product Launch Model

For the new product launch, NorthStar's analysts have gathered the following information from market research and historical data:

  • Demand: Normal distribution with mean = 10,000 units and SD = 2,000 units
  • Unit Price: Uniform distribution between $45 and $65 (varies by sales channel)
  • Unit Cost: Triangular distribution with min = $20, mode = $28, max = $38
  • Fixed Cost: $150,000 (known with certainty)
Profit Equation
📊 Excel: =(Price-Cost)*Demand - FixedCost
where Price is the unit selling price, Cost is the unit production cost, Demand is the number of units sold, and FixedCost is the fixed overhead.
Normal Distribution — Random Generation
📊 Excel: =NORM.INV(RAND(), 10000, 2000)
Generates a random demand value from a Normal distribution with mean μ = 10,000 and standard deviation σ = 2,000.
Uniform Distribution — Random Generation
📊 Excel: =45 + 20*RAND()
Generates a random price between a = $45 and b = $65. Every value in the range is equally likely.
Triangular Distribution — Expected Value
📊 Excel: =(20+28+38)/3
where a = $20 (minimum), c = $28 (mode), and b = $38 (maximum). For random generation in Excel, use an inverse CDF approach with RAND().

Steps 2–4: Generate, Calculate, Repeat

In Excel, each row represents one simulation trial. You generate random values for demand, price, and cost, then compute profit for that row. Copy the formulas down for 1,000 or more rows — each row is an independent "what if" scenario drawn at random from the input distributions.

Step 5: Analyze the Output

The result is a distribution of 1,000 profit values. From this distribution you can extract key decision metrics: the mean profit, the probability of profit exceeding zero, the 5th percentile (worst-case threshold), and the 95th percentile (best-case threshold).

✎ Worked Example: Setting Up NorthStar's Simulation in Excel
1
In column A, generate random demand:
=NORM.INV(RAND(), 10000, 2000)
2
In column B, generate random unit price:
=45 + 20*RAND()
3
In column C, generate random unit cost (triangular):
=20 + (38-20) * (RAND())^(1/2) * ... [use helper formula]
4
In column D, compute profit:
=(B1-C1)*A1 - 150000
5
Copy rows 1–4 down to row 1000. Then compute summary statistics:
Mean: =AVERAGE(D1:D1000)   P(Profit>0): =COUNTIF(D1:D1000,">0")/1000
5th %ile: =PERCENTILE.INC(D1:D1000, 0.05)   95th %ile: =PERCENTILE.INC(D1:D1000, 0.95)
NorthStar Product Launch — Monte Carlo Simulation
Click to run 1,000 iterations
🎮
Monte Carlo Casino Build and run your own simulations to make business decisions under uncertainty

8.3 Interpreting Simulation Results

Running the simulation is only half the work. The real value lies in interpreting the output distribution to support business decisions. Three key techniques help extract actionable insights: summary statistics, sensitivity analysis, and risk metrics.

Summary Statistics

The mean of the simulated profit distribution represents the expected profit under uncertainty. For NorthStar, the simulation yields a mean profit of approximately $142,000. But the mean alone can be misleading — a business that earns $1 million half the time and loses $700,000 the other half has a positive expected value but extremely high risk.

Sensitivity Analysis: The Tornado Chart

Which uncertain input drives the most variation in profit? A tornado chart answers this question by showing how much the output swings when each input moves from its low to high value while the others stay at their base case. The widest bar represents the most influential input. For NorthStar, demand has the largest impact because it multiplies against the entire margin, followed by unit price, and then unit cost.

Sensitivity Analysis — Tornado Chart

Value at Risk (VaR)

Value at Risk answers the question: "What is the worst loss we could expect at a given confidence level?" The 5th percentile of the profit distribution serves as the VaR at 95% confidence. For NorthStar, the 5th percentile is approximately −$48,000, meaning there is a 5% chance that the product launch will lose $48,000 or more.

Probability of Loss

The simulation shows that approximately 87% of trials produce a positive profit, meaning there is about a 13% chance of loss. Management must decide whether this level of risk is acceptable given the expected upside.

🏪 NorthStar Enterprises

After reviewing the simulation results, NorthStar's executive team sees that the product launch has an 87% probability of profit, with an expected profit of $142,000. However, the 5th percentile shows a potential loss of $48,000 — the tail risk. The tornado chart reveals that demand uncertainty is the primary driver of risk.

Management decides to proceed with the launch but invests in additional market research to narrow the demand estimate, which would reduce the simulation's spread and improve confidence in the decision.

✓ Check Your Understanding
In a Monte Carlo simulation, the 5th percentile of the profit distribution represents:
The most likely profit outcome
The value below which only 5% of outcomes fall (worst-case threshold)
The average profit across all simulations
The probability of earning a profit
💡 Key Takeaway

Monte Carlo simulation does not remove uncertainty — it quantifies it. By translating vague worries about "what could happen" into specific probabilities and dollar amounts, simulation empowers decision makers to choose how much risk is acceptable. The technique is only as good as the input distributions: garbage in, garbage out. Always validate your assumptions with data and domain expertise.

8.4 Chapter Summary

Monte Carlo simulation is one of the most versatile tools in the business analyst's toolkit. It transforms complex, multi-variable uncertainty into a concrete distribution of outcomes that can be measured, visualized, and acted upon. From product launch decisions to capital budgeting to supply chain risk assessment, simulation provides the quantitative foundation for smarter decision-making under uncertainty.

💡 Chapter 8 Summary

Why Simulate: When business problems involve multiple uncertain inputs and no closed-form solution exists, Monte Carlo simulation uses random sampling to estimate the full distribution of outcomes.

Building a Simulation: Define input distributions (Normal, Uniform, Triangular), generate random values, calculate the output, repeat thousands of times, and analyze the resulting distribution.

Interpreting Results: Use summary statistics (mean, percentiles), sensitivity analysis (tornado charts), and risk metrics (VaR, probability of loss) to translate simulation output into actionable business insight.

📋 Chapter 8 — Formula Reference
Concept Formula Excel Function
Profit
=(Price-Cost)*Demand-Fixed
Normal Random
=NORM.INV(RAND(),μ,σ)
Uniform Random
=a+(b-a)*RAND()
Triangular E(X)
=(a+c+b)/3
Mean Profit
=AVERAGE(profit_range)
Percentile (VaR)
=PERCENTILE.INC(range,k)
Course Complete
Congratulations! You've completed STATS300: Advanced Business Statistics.