A single completion date is a comforting thing to write in a contract. It is also, almost always, wrong — and worse, it throws away the one piece of information a decision-maker actually needs: how wrong it might be. The case for quantitative risk analysis isn't really about producing a bigger, safer number. It's about refusing to discard the distribution.
Plan with single best-guess durations and you commit a subtle statistical error long before any risk shows up. Because schedules converge — parallel paths merge, and the late one wins — the average of the outcomes is not the outcome of the averages. Feed in "most likely" durations and you get a date the project has well under a 50% chance of hitting. Rob Brown, writing for Analytica, calls the habit behind this inattention risk: point-estimate thinking "masks significant risk" because it never shows the spread of futures that remain live (Analytica, "The Value of Knowing What You Don't Know").
Brown's illustration is a hotel-guarantee valuation, and the numbers make the point better than any schedule metaphor. Valued with single-point estimates, a performance guarantee was worth $1.5 million. Modelled properly — with Monte Carlo over the uncertain inputs — its expected value was $3.4 million, more than double. As Brown puts it, "this information is lost by using single point analysis." The $1.9M gap even has a name: the expected value of including uncertainty — value that was always there, simply invisible to a spreadsheet of best guesses.
Schedules behave the same way. The deterministic date sits in the optimistic tail of a distribution the plan never draws; the contingency you actually need lives in the part of the range the single number erased.
Keep the whole distribution and three decisions get easier at once:
A common objection is that risk analysis just makes every date later and every meeting gloomier. It doesn't. Brown's guarantee was worth more once uncertainty was modelled, not less; the distribution surfaced upside as readily as downside. In schedule terms, quantifying the range is what lets you defend a date, size a hedge, and spot the opportunities — the activities where a little acceleration buys a lot of confidence — that a deterministic plan flattens into nothing.
RunMonteCarlo turns a P6 or MS Project schedule into that distribution in a couple of minutes — completion curve, P-values and confidence bands included, so the range is something you can point at rather than argue about.
RunMonteCarlo turns your schedule into a completion curve with P50/P80 and confidence bands — the contingency story in one picture. Free tier included; your schedule never leaves your machine.
Launch RunMonteCarlo →Latin Hypercube's edge is largest when a few inputs dominate the output. When many variables contribute roughly equally, their independent errors partly cancel on their own, and the stratification advantage narrows — David Vose's nine-input test is the classic demonstration that with enough equally-weighted drivers, LHS and plain MC converge (Risk Analysis: A Quantitative Guide).
Here is the thing: schedule risk is almost never the equal-weight case. A project's completion distribution is driven by the handful of activities on and near the critical path. Everything else has float and contributes almost nothing to the finish date. That is the exact regime — a few dominant drivers — where Latin Hypercube pays the most. The property that makes hand methods fail (only a few paths matter, and they shift under risk — see merge bias) is the same property that makes LHS shine.
They should be — trades share weather, crews and productivity, and pretending otherwise understates the tail. A fair worry is whether stratifying each input separately breaks the correlation structure. It doesn't. Latin Hypercube composes cleanly with correlation: you stratify the marginals, then impose the dependency by reordering the samples to hit a target rank correlation — the Iman & Conover method (1982) — or, equivalently, drive the strata through a copula. RunMonteCarlo uses a Gaussian copula for correlation and still stratifies underneath it, so you get correlated and well-covered durations at once. Details are on The Maths.
Effectively no. The one-time cost is turning a uniform draw into a value on your distribution — the inverse CDF. For the Normal that has no closed form, but Acklam's rational approximation computes it to machine precision in a few multiplications, so an LHS draw costs about the same as a plain one. And in a schedule model the sampling is a rounding error next to the real work: re-solving the CPM network over thousands of iterations. You get the variance reduction essentially for free. RunMonteCarlo already cites Acklam's inverse-normal in The Maths for precisely this reason.
Latin Hypercube isn't a buried expert option in RunMonteCarlo — it's how the engine samples by default, and it's the same seeded, reproducible pipeline the rest of the tool runs on: