Monte Carlo simulation for decisions: what it adds to a spreadsheet
The essentials in 30 seconds
- Monte Carlo simulation replaces each input value with a range and runs thousands of draws.
- In the illustrative case, the single-point profit is 8,000 while the mean of 10,000 runs is 29,258.
- The chance of a loss is about 30 percent, a figure the single-point model hides.
- Five runs with different seeds return loss shares from 28.7 to 30.0 percent.
- The EQOS Asia decision engine takes a deterministic path and returns the same score for the same inputs (EQOS Asia method page, as of October 2026).
A spreadsheet holds one value per cell, so it answers one question: what happens if every input lands where the plan says. Monte Carlo simulation asks a wider one: what happens across thousands of plausible combinations. This guide explains the method on an illustrative case, with the arithmetic and the settings shown, and places it among the decision making frameworks.
On this page
Compare the EQOS Asia offers and request a 30-minute pre-diagnostic.
What does Monte Carlo simulation add to a one-value spreadsheet?
A spreadsheet forecast puts one number in each input cell and returns one number at the bottom. That number rests on a hidden claim: every input lands exactly on its estimate. Monte Carlo simulation replaces each uncertain input with a range, draws random values from the ranges, and recomputes the sheet thousands of times.
The output is a distribution instead of a point. A reader sees the median, the spread, and the share of runs that end in a loss. The same formulas remain in place, so the team keeps its model and gains a view of the risk around it, with a number for the chance of missing the target.
How does a Monte Carlo simulation work, step by step?
Five steps run the method. Build the model in a sheet or a short script. Give each uncertain input a range and a shape. Draw one value per input and compute the result. Repeat the draw many times, often 10,000. Then read the distribution of the results, not the single output of any one run. The full cycle fits in a few seconds on an ordinary laptop.
Because each draw is independent, the method is simple to check. A fixed seed for the random generator makes a run reproducible, so a colleague who repeats it obtains the same figures. Record the seed, the number of draws, and the ranges next to the result, so the run can be replayed later.
Reading the output is a skill of its own. Monte Carlo simulation explained for decision makers covers percentiles, the chance of missing a budget, the inputs that drive the spread, and five questions that test the quality of any simulated result before a board relies on it.
How do you choose ranges for the uncertain inputs of a simulation?
A triangular range needs three numbers: the lowest plausible value, the most likely value, and the highest plausible value. It is easy to elicit from an expert and handles skewed inputs well. A uniform range needs only a minimum and a maximum and suits an input with no favourite value. Each range is written on the sheet beside its source.
Ask for the range before the most likely value, since people who name the central estimate first tend to set the range too tight. Note the source of each range. In the example below, units sold follow a triangular range from 6,000 through 9,000 to 15,000, and the margin per unit is uniform between 18 and 26.
Most finance teams already work in a spreadsheet, and that is a fine place to begin. A guide to Monte Carlo simulation in Excel shows the layout, the formulas, and how much the results move at each recalculation for a service line example, so that a finance team can build its first simulation without an add-in.
What does a worked Monte Carlo example return for a product line?
Take a product line, all figures illustrative. Profit equals units times margin per unit, minus fixed costs of 190,000. The single-point sheet uses 9,000 units and a margin of 22, so profit is 198,000 minus 190,000, or 8,000. The case looks positive but thin.
A run of 10,000 draws with seed 2026 returns a mean profit of 29,258 and a median of 24,635. The 10th percentile is minus 28,464 and the 90th percentile is 93,463. Profit is negative in 30.0 percent of runs. The sheet showed a small gain, and the simulation shows a three in ten chance of a loss.
How many runs are enough for a Monte Carlo result?
Enough runs mean that the answer stops moving. With 10,000 draws, five runs with different seeds returned a loss share between 28.7 and 30.0 percent, so the figure is stable to about one point. Repeat the run at several sizes and several seeds, and stop increasing the draws when the headline figures stop moving.
Precision of the model matters more than precision of the simulation. A tight loss share of 30.0 percent still rests on the ranges chosen for the inputs. Spend the effort on the ranges and their sources, and report the loss share as about 30 percent instead of an exact value. The sensitivity of the profit to each input then shows where more evidence pays.
Which limits of Monte Carlo simulation deserve attention?
A simulation returns what the ranges contain. If the ranges are guesses, the distribution is a tidy picture of guesses. Inputs that move together, such as sales volume and price in a weak market, must be linked in the model, or the spread of outcomes is understated. A short review of the correlations takes minutes.
Rare events outside the ranges stay invisible, so pair the simulation with a worst-case scenario written by hand. Present the distribution with its assumptions in one table. A clear list of ranges and sources lets a reviewer challenge the inputs and not just admire the chart.
How does the EQOS Asia decision engine relate to Monte Carlo simulation?
Monte Carlo is a probabilistic method: two runs with different seeds return slightly different figures. The EQOS Asia decision engine takes the deterministic path. Its Codex Universalis core runs pure, verifiable, reproducible mathematics, and every verdict can be recomputed identically (EQOS Asia method page, as of October 2026).
The Prediction pillar projects 3 trajectory scenarios with deterministic probabilities and dated tipping thresholds. The two approaches complement each other: a simulation sizes the spread of a single model, and the engine scores the situation on 9 dimensions. A reader can try a first score in the simulator, and the 30-minute pre-diagnostic, scheduled in Singapore time (SGT), applies the engine to a live decision.
See also in this guide
Frequently asked questions
What is Monte Carlo simulation in simple terms?
Monte Carlo simulation recomputes a model thousands of times, each time drawing random values for the uncertain inputs from stated ranges. The collected results form a distribution that shows the typical outcome, the spread, and the chance of ending above or below a target.
Does a Monte Carlo simulation need special software?
A spreadsheet with a random-number function can run a simple simulation, and a short script in a common programming language runs a larger one with a fixed seed. The method matters more than the tool: stated ranges, a documented seed, and a reviewed model.
How does Monte Carlo differ from scenario analysis?
Scenario analysis writes three coherent stories by hand. Monte Carlo samples thousands of combinations from ranges and reports the spread. The first explains why outcomes differ, and the second measures how often each outcome appears. Many teams use both on the same decision.
Is the EQOS Asia engine a Monte Carlo tool?
The EQOS Asia engine runs a deterministic core, so the same inputs return the same score on every run, and it returns deterministic probabilities rather than sampled ones (EQOS Asia method page, as of October 2026). A simulation can size the spread of a single model beside it.
Test your knowledge
1. What does Monte Carlo simulation return?
- A single forecast
- A distribution of outcomes
- A ranked list of criteria
Show the answer
A distribution of outcomes.
2. What is the single-point profit in the example?
- 8,000
- 29,258
- 24,635
Show the answer
8,000.
3. In what share of runs is profit negative in the example?
- 5.0 percent
- 50 percent
- 30.0 percent
Show the answer
30.0 percent.
4. Why record the seed of the random generator?
- It sets the number of inputs
- It makes a run reproducible
- It raises the median
Show the answer
It makes a run reproducible.
5. What does the EQOS Asia core return for the same inputs?
- A slightly different score
- A range of scores
- The same score
Show the answer
The same score.
About the author
Stéphane Jambu
President, EQOS Forecast Asia
Stéphane Jambu leads EQOS Forecast Asia and works as a consultant in e-commerce, search visibility, and AI. LinkedIn profile
Sources: EQOS Asia method page (eqos-asia.com/method.html), simulator page (eqos-asia.com/simulator.html), and contact page (eqos-asia.com/contact.html), consulted on 21 October 2026. Page updated on 21 October 2026.