Monte Carlo forecasting in Jira, explained simply

Monte Carlo sounds like statistics homework. It is closer to rolling dice with your own history, many times, and looking at where the results land.

The idea in one paragraph

You do not know how fast the next weeks will go. You do know how fast the last weeks went. Monte Carlo builds thousands of possible futures out of those real past days, then tells you how often each finish date comes up. Dates that come up often are likely. Dates that almost never come up are risky promises.

A worked example

Say the last 84 days show a team finishing 0, 1, 2 or 3 items a day, about 1.5 on average. Forty items remain. Dividing gives roughly 27 days, but that average hides the spread. Across 10,000 simulated futures, half finish within about 27 days, 85% within about 31, and 95% within about 34. The last two numbers are the ones worth promising.

Why it beats story points

• No estimation meetings: it uses what the team finished, not what it guessed.
• It shows risk directly: the gap between the 50% and 95% dates is your uncertainty.
• It updates itself: every finished item sharpens the next forecast.

Common mistakes

• Mixing epics with stories, which double-counts work.
• Using a history window that includes a different team or a holiday season only.
• Promising the 50% date. Half of all futures miss it.

Try it on your own project

Clearlane Forecast runs 10,000 simulations inside Jira and shows the likely, safe and rock-solid dates on one screen, with a “what if” to test scope changes.