Decision & Risk Simulator teaching model

Seeded Monte Carlo arithmetic, criterion-specific comparisons, linear sensitivity checks, and expected-monetary-value decision trees.

Scope: outputs are hypothetical results from assumptions you enter. They are not forecasts, guarantees, financial advice, or a substitute for validating costs, probabilities, dependencies, taxes, fees, and real-world constraints.
Ready.

Options & assumptions

Allowed: variable names, numbers, + − × ÷ ^, parentheses, and min(...), max(...), abs(...), sqrt(...). No JavaScript execution.

Variables

Simulated outcome summary

Distribution view

Sample percentiles use linear interpolation between ordered simulated outcomes. Sample extrema are not guaranteed distribution bounds.

The first row is only the leader for the selected criterion—not a universal “best” choice.
OptionMeanMedianStd devP(loss)P(target)P10P90Interpretation

What-if locks

Locking a variable makes it constant for the next run. Correlation is then undefined for that constant input and shown as n/a.

Linear sensitivity screen

Pearson r describes sample linear association between each input and the modeled payoff. It does not establish causation and can miss nonlinear effects.

Decision tree rollback

Chance nodes require outgoing probabilities to total 1. Decision nodes use maximum expected monetary value only—a risk-neutral criterion.

Selected node

Decision-risk report

Hypothetical mathematical illustration only. Results depend entirely on modeled assumptions and omit any real-world factor not explicitly represented.

Option summary

OptionMeanMedianStd devP(loss)P(target)P10P90

Assumptions & formulas

Decision-tree scope

Methodology & limits

  • Randomness: a seeded, deterministic browser PRNG makes the same saved model + same seed reproducible within this app version. It is not cryptographic randomness.
  • Normal: Box–Muller sampling with user-supplied mean and non-negative standard deviation. Normal tails are unbounded; the app does not silently truncate physically impossible values.
  • Triangular: requires min ≤ mode ≤ max and uses inverse-CDF sampling.
  • Bernoulli/discrete: probabilities are validated. Discrete outcome probabilities must sum to 100% (within rounding tolerance).
  • Percentiles: empirical quantiles from simulated outcomes; they are not promises, confidence guarantees, or worst-case bounds.
  • Sensitivity: Pearson product-moment correlation is a linear association screen, not a causal proof or complete global sensitivity analysis.
  • Decision tree: chance probabilities must sum to 1. A decision node selecting the largest expected monetary value assumes risk-neutral EMV; other risk preferences may choose differently.
  • Dependencies: variables are sampled independently unless dependence is built into the formula itself. Correlated inputs are not modeled.