Monte Carlo risk analysis

What it does

Takes a live schedule update and re-runs its schedule logic hundreds or thousands of times, each time varying the remaining duration of every unfinished activity within a range you set. The spread of finish dates that results shows how likely the project is to finish by any given date, which activities most often sit on the chain that sets the finish, and which durations the finish is most sensitive to.

What it helps you achieve

• Replace a single finish date with a range, and say how confident you are in it (for example, an 80% chance of finishing by a date).
• Find out the chance of meeting a contractual or target date.
• Focus attention on the few activities that most often drive the finish and that the finish is most sensitive to.

How to run it

Where to find it: Project page → Monte Carlo risk analysis

1. Open the project and choose Monte Carlo risk analysis.
2. 1. Choose the schedule update. Use a live, in-progress one: only unfinished work is varied.
3. 2. Set the uncertainty: the optimistic and pessimistic durations as a percentage of each activity's planned remaining duration (for example 90% and 125%), the shape of the distribution, and optionally a target finish date.
4. 3. Leave the iterations and seed as they are, or change them. The same seed and settings always give the same answer.
5. Choose Run Analysis.
6. Read the planned finish against the P50 and P80 finish dates, the chance of meeting your target, the two charts, and the tables of the activities that most often drive the finish and that it is most sensitive to.

What you get

• P50, P80 and P90 finish dates, and how many days each sits from the planned finish.
• The probability of finishing on or before a target date you choose.
• A histogram of simulated finish dates and a confidence curve.
• For each activity, how often it drove the finish and how closely its duration tracked the finish date.
• A saved result you can reopen, and a place in the Word delay report.

Good to know

• With every duration at its planned value the simulation reproduces the schedule recalculation engine's finish exactly, so the range is centred on a schedule you can already trust.
• Change the ranges and run again to see how much the answer depends on them. That sensitivity is itself a finding.
• Large schedules stop after a time limit and say how many iterations they completed.

Frequently asked questions

What do P50 and P80 mean?

P50 is the finish date that half of the simulated runs beat; P80 is the date that 80% of runs beat. A P80 date is a more cautious commitment than the planned finish or the P50.

Why does the answer change when I change the ranges?

Because the ranges are the assumption the whole analysis rests on. Wider pessimistic ranges push the P80 later. Seeing how far it moves shows how much to trust it.

Go deeper

• What total float tells you about risk →

Terms used here

• Critical Path
• Total Float