Docs / Risk
Reading simulation output
S-curves, finish histograms, criticality, and tornado sensitivity — what each one tells you, and what it doesn't.
The S-curve
The cumulative probability of finishing by each date. Use it for commitments (read off P80) and for honesty about the deterministic date (it usually sits near P20–P40, i.e. you'd miss it most of the time). It does not tell you why — that's the next two charts.
The finish histogram
The same data as the S-curve, shown as a distribution. A long right tail means a few ugly futures are dragging your P80 out — worth finding which tasks cause them. Two humps usually mean a binary risk (a task that either happens or doesn't) splitting your futures into two families.
Criticality index
For every task: the share of iterations in which it sat on the critical path. A task that's critical in 84% of runs is your real bottleneck even if today's deterministic path skips it. Tasks with low criticality but high uncertainty are usually safe to ignore — their variation is absorbed by slack.
Tornado sensitivity
Which task's uncertainty moves the finish date most, ranked. This is your work list: tighten the estimate (or de-risk the work) at the top of the tornado and re-run. The chart tells you where buying information pays; it doesn't tell you how — that's engineering.
What none of this tells you
The simulation only knows the risks you modelled. A risk that isn't in the plan — a scope change, a vendor you didn't model — isn't in the distribution. Treat P80 as "80% confident, given the plan as drawn", and keep the plan honest.