A Monte Carlo delivery forecast uses the team's historical throughput (items finished per week, sprint or day) to simulate thousands of possible futures. Each simulation draws real past periods until the work is complete. The result is a distribution: how many periods the batch takes with a 50%, 85% and 95% chance, or how many items fit by a date. There is no need to estimate the size of each item.
How to use
- Paste the number of items finished in each of the last periods, ideally 8 to 12.
- Choose the history period: weeks, sprints or days.
- Enter the items remaining to see when the batch will be done, the periods until the date to see how many items fit, or both.
- Read the percentiles: 85% is the usual basis for agreeing a date or scope; 50% is a coin toss.
- Rerun the forecast every period with the updated history.
What each field asks for
| Field | What to enter |
|---|---|
| Items finished per period | One number per period, most recent first, separated by commas, spaces or lines (up to 52). |
| History period | The forecast uses the same unit. |
| Items remaining | To answer when the batch will be done. |
| Periods until the date | To answer how many items fit by then. |
Example
A team finished 6, 4, 7, 3, 5, 8, 2, 6, 5, 4, 7 and 5 items over the last 12 weeks, an average of 5.17 a week. There are 40 items left. Dividing by the average would give 7.7 weeks, but the simulation shows a 50% chance in 8 weeks, 85% in 9 and 95% in 10. For the other question, in 6 weeks the team finishes 31 items or more with a 50% chance, 27 or more with 85% and 24 or more with 95%.
When to use
- When someone asks "when will it be done?" and the answer needs a confidence level.
- To decide how much scope fits by a fixed date, such as a launch or an event.
- To replace item-by-item estimates with a forecast based on what the team actually delivers.
- To track, period by period, whether a commitment is still realistic.
Frequently asked questions
What is a Monte Carlo simulation in delivery forecasting?
It means drawing, many times and with replacement, the throughput of real periods from the history and seeing how long each simulated future takes to finish the work. The tool runs 5,000 simulations and summarises the result in percentiles.
How many periods of history should I use?
Between 8 and 12 is usually enough, as long as they represent the current team and type of work. Fewer than 5 makes the forecast fragile; very old or unusual periods, such as holidays, distort the result.
Why use 85% rather than the average?
Dividing the items by the average gives a date that is met roughly half the time. 85% means that in 85 out of 100 simulated futures the work is done by that period: a more honest basis for a commitment.
Do I need to estimate the size of each item?
No. The forecast counts items and assumes the remaining ones are of a similar size to those in the history. If one item is much larger, split it before counting.
Does the forecast change if I reload the page?
No. The simulation uses a number generator with a fixed seed: the same input always gives the same result, so you can compare forecasts and share the link.
How does it relate to Little's Law?
Little's Law links averages of work in progress, throughput and cycle time, and explains the system. The Monte Carlo simulation answers a date or scope question with a probability. Both rely on throughput.