01
Set the ranges
Choose what could realistically happen to price, demand, costs and other drivers.
Fuel price−10% to +30%
Knowledge for finance
One forecast is one answer.
Monte Carlo shows the range.
You provide the financial model, the assumptions and their possible ranges. Scenario Builder tests thousands of random combinations and shows what they mean for profit, cash flow or any other result.
See the four stepsOne simulation Run 4,281 / 10,000
Sales volume+2.4%
Selling price−1.1%
Fuel price+18.6%
Financial formulas recalculate
Operating profit€0.94m
Save this outcome. Pick again. Repeat 10,000 times.
Monte Carlo in plain words
You give the range of assumptions.
The app picks the numbers.
Every pick creates one possible financial future. Repeating this many times produces a range of outcomes instead of one fixed forecast.
02
Pick random values
The app picks one value for every uncertain input, within the range you provided.
−4%+17%+3%+26%
03
Recalculate the model
Your existing formulas calculate revenue, costs, profit and cash flow for that combination.
Profit = Revenue − Costs
04
Repeat thousands of times
The app sorts all results and counts how often the budget, break-even point or target is reached.
10,000possible outcomes
The formulas stay the same.
Only the uncertain assumptions change from one simulation to the next.
Correlation
Some risks belong together.
Correlation tells the app which random values should appear together in the same scenario. It does not make an input larger and it does not apply a movement twice.
Positive correlation
Fuel and materials often rise together
FuelMaterials
Lowpaired withLow
Averagepaired withAverage
Highoften paired withHigh
Bad cost movements are more likely to arrive together.
Negative correlation
Price and volume can partly offset
PriceVolume
Lowoften paired withHigh
Averagepaired withAverage
Highoften paired withLow
A stronger movement in one input can be partly balanced by the other.
What happens in the background?
The app correlates the full set, not one number.
- 1Generate thousands of random values for each uncertain input.
- 2Sort every input’s values from low to high.
- 3Rearrange which values are paired in the same scenarios.
- 4Check that the full set approximately matches the requested correlation.
Nothing is doubled.
Fuel set between −10% and +30% stays between −10% and +30%. Correlation only changes
which material-cost value sits beside it.
Reading the outcome
From thousands of runs
to one useful picture.
The outcomes are sorted from poor to strong. Percentiles show where results landed.
Example: first-year operating profit€0.3m – €1.8m
Chance of beating budget36%
Budget €1.2m
P5€0.3m
P50€1.0m
P95€1.8m
P5 · Downside5% of simulated outcomes were lower and 95% were higher.
P50 · MedianHalf were lower and half were higher. It is not necessarily the most frequent result.
P95 · Upside95% of simulated outcomes were lower and 5% were higher.
“
The median simulated profit is €1.0 million. The P5 downside is €0.3 million, while the P95 upside is €1.8 million. The €1.2 million budget was reached in 36% of scenarios.
A range is not a guarantee.
The result only describes the scenarios created from your formulas, ranges and correlations.
Better assumptions produce more useful outcomes.
Why percentile lines do not always add up.
A P50 for revenue and a P50 for cost can come from different simulations, so subtracting
those two marginal medians does not necessarily equal the P50 for profit. In the report,
the Tree and component reconciliation use one real simulation close to the selected
headline percentile. Those lines come from the same scenario and therefore reconcile.
Driver ranking is a screening view.
It moves one input at a time while the other uncertain inputs stay on plan. This is useful
for prioritising assumption reviews, but it is not a variance contribution and does not
measure interactions or combined correlated risk.
Why the setup works
Known formulas first.
Statistics where they help.
Multicollinearity
Correlated inputs do not stop the simulation
Fuel and materials may move together. This makes it harder to separate their individual effects, but their combined effect is still calculated correctly.
Why? The financial formulas are already known. The app does not need
to estimate them from the data.
Beta
Beta is not needed or set up in this Monte Carlo
Beta is useful when a model must estimate how strongly one variable affects another. Scenario Builder does not estimate that relationship.
Revenue = Volume × Price
The relationship is already defined. The app changes the inputs and
uses the formula directly, so no regression or market beta is required.
What finance can do with it
Turn uncertainty into decisions.
01See a realistic range
Move beyond one-point forecasts.
02Measure target probability
See the chance of meeting budget.
03Understand downside
Estimate loss and covenant risk.
04Find key assumptions
Focus management attention.
05Test decisions
Compare price, volume and cost plans.
06Explain the result
Connect outcomes to business drivers.
Security and data handling
What is protected today.
And what is not claimed.
Scenario Builder uses invited accounts, server-side organisation checks and encrypted cloud services. This summary describes the current production setup; it does not claim that Scenario Builder itself is SOC 2 or ISO 27001 certified.
Access
Invitation-based and organisation-scoped
Every report, workbook download, run and delete request is checked on the server against the signed-in user’s organisation.
Sessions use Secure, HTTP-only, SameSite=Strict cookies with a five-day expiry.
Encryption
Encrypted in transit and at rest
HTTPS protects traffic in transit. Google Cloud encrypts stored customer content at the storage layer using AES encryption with Google-managed keys.
Processing
Model processing in the EU
Workbook and report storage, database records and Monte Carlo processing use the project’s Google Cloud EU configuration, including europe-west1 compute.
The web application is hosted by Vercel and securely forwards requests to the model services.
Use of data
No advertising or model-data mining
The application contains no third-party advertising or behavioural analytics scripts. Uploaded workbooks are used to produce the requested reports and exports.
Current retention limitation.
Scenario files can be deleted in Studio. Generated reports from earlier runs may remain,
and there is not yet a fixed automatic retention period. Do not interpret scenario deletion
as a complete account-wide erasure request; contact the administrator when full removal is required.
Current subprocessors
| Provider | Purpose | Data involved | Location / control |
|---|---|---|---|
| Google Cloud / Firebase | Authentication, database, workbook and report storage, Monte Carlo processing | Email and organisation metadata, uploaded workbooks, run records and generated reports | Core storage and model processing configured in the EU; Google-managed encryption keys |
| Vercel | Website hosting and server-side application requests | Encrypted request, session and transient upload traffic needed to operate the service | Handled under Vercel’s platform controls and data-processing terms |
Provider lists can change. See the Vercel Trust Center ↗ and Vercel DPA ↗. Last updated 22 July 2026.
The complete idea
You provide the logic.
Monte Carlo tests the possibilities.
Scenario Builder does not promise one exact future. It shows the possible futures, how likely they are, and which assumptions deserve attention.