Many Worlds

Scenario planning across a million futures

← dinand.com
About this tool

Scenario planning that tests every response against the same futures

Many Worlds plays a business scenario out hundreds of thousands of times. Each simulated future draws its own mix of demand, prices, costs and shocks, so the spread of outcomes shows how wide the plausible range runs before anyone commits money.

The engine then runs every possible response through all of those futures. It compares each set of moves against doing nothing on identical draws and recommends the one that holds up best for the risk appetite set in the side panel.

The five library scenarios carry illustrative figures. For a client, replace them with real numbers under Assumptions or paste historical data to fit the ranges.

Recommended responseDoing nothingEach colour keeps its meaning in every chart on the page.
  1. Pick a scenarioChoose one from the library on the left, or describe a client situation in plain words.
  2. The engine plays out the futuresEvery driver is drawn from its range, and shocks land at random times with random force.
  3. Every response meets the same futuresEach combination of moves is scored on identical draws, including moves held in reserve until trouble starts. Any gap between two policies therefore comes from the moves themselves.
  4. The call gets stress-testedThe winner is rerun with one assumption changed at a time before it reaches the top of the page.
Policy
A combination of moves, where any move can start now or wait in reserve as an option.
Option
A move prepared now for 15% of its cost and exercised only if its trigger fires.
Worst tenth
The average result across the worst 10% of futures, a gauge of how bad a bad outcome gets.
Risk-adjusted score
Expected value blended with the worst tenth, weighted by the risk posture you choose.
Unacceptable outcome
A result below the floor set in the side panel, which by default is the level doing nothing misses in a fifth of futures.
Recommended responseRefining

Running the first simulation.

Reasoning

Why this is the call

Robustness

How sturdy the call is

Plan

The plan, quarter by quarter

Committed moves start this quarter, while an option gets prepared now and is exercised only if its trigger fires.

Moves

Value of each move

Each bar shows how far the risk-adjusted score moves when you add a move to the recommended set or take it out. Every comparison runs on the same futures.

Outcomes

Where the outcomes land

Failure modes

When the call is wrong

The engine searches the futures for the conditions that best explain where the recommendation loses.

Stress tests

How the call fares when assumptions are wrong

Exposure

Exposure that remains

The average outcome under the recommended response when each uncertainty sits at either end of its range.

Budget

Return on more budget

The best score the search found at each level of upfront spend.

Search

Every policy tested

Each dot is one combination of moves, committed or held as options. A dot sits higher when its expected value is larger and further right when its worst tenth of futures hurts less.

Finalists

The finalists, side by side

Analyst

Ask Claude

Claude reads the scenario and the results, and reruns the simulation itself to test a what-if. It uses your own Claude account.

Inputs

Assumptions

The engine uses every input listed here, and editing any value reruns the simulation. Each range runs from a plausible low to a plausible high, with its peak at the most likely value.

Fit ranges and correlations from data

Paste a table with a header row and one column per driver, in the driver's own unit such as yearly % change. Each fitted range spans the 5th to the 95th percentile and peaks at the median. Correlations come from the matched columns.