Quantitative Risk & Contingency Analysis
Use this workbench when a bounded project or program needs a defensible distribution of final cost, completion duration, or both. It connects an unrisked baseline, base-estimate uncertainty, discrete threats and opportunities, dependencies, scenarios, evidence, and an explicit contingency policy to a reproducible Monte Carlo run.
The workbench can answer what P50, P80, P90, or another declared percentile means under the recorded model; how much contingency or schedule reserve separates that percentile from the unrisked baseline; the modeled confidence at a human-entered budget or deadline; and which inputs are associated with the upper tail. It cannot guarantee an outcome, approve funds, replace a resource-loaded schedule, or manufacture precision from weak evidence.
Core boundary
Base uncertainty exists even when no named event occurs. A discrete event has a probability of occurrence and a non-negative impact magnitude: threats add and opportunities subtract. Existing contingency, escalation, allowances, and management reserve remain separate comparison inputs; none belongs inside the unrisked baseline. Inherent and residual profiles are independent claims, and residual exposure requires an explicit response, owner, status, and rationale.
Use the qualitative Risk & Assumption Stress Test when credible numerical ranges are unavailable; a Decision Tree for sequential choices and learning; Scenario Planning for deep uncertainty; Project Viability for cash-flow sustainability; or Portfolio & Resource Allocation for choosing a feasible mix of initiatives.
Supportable ranges can form a cost or duration distribution
A percentile guarantees delivery or authorizes reserve
Evidence is mainly qualitative or ordinal
Ordinal exposure is a monetary contingency
Choices and information arrive sequentially
A tree is an aggregate project-risk distribution
Deep uncertainty makes precise probabilities misleading
A plausible future has an objective probability
Begin with selection and framing, then progress through the pages in order. ## Interpretation discipline
Keep inputs, derived results, and human decisions visibly separate. Every material number needs linked evidence or an explicit assumption flag and basis. Unknown is not zero. A percentile is conditional on this model, scenario, seed, engine version, and source version; it is neither a guarantee nor an approval. Any material model edit makes an earlier result stale.