Chance of done
October 2026 · forecasting, flow
You have work left to do and a team with a record of how much it finishes each week. Instead of guessing one date, this page replays the team’s past weeks in random order, five thousand times, and counts how many of those futures are done by each coming week.
Reading the forecast
The 50% date is a coin toss: half the futures finish later. Plan on the 85% date, and say the chance out loud when you share it, the way a forecaster says “70% chance of rain”.
Storms
Your history only contains the kinds of weeks you’ve had. Turn up storm season and the page starts dropping in rough stretches of one to four weeks with nothing done: an outage, a reorg, someone leaving. Watch both dates. They both move, and the 85% date moves further, because the late futures are the ones a storm hit.
Looking back
The example team slowed down thirteen weeks ago. Look back over all 26 weeks and the forecast leans on a faster team that no longer exists. Look back over the last four and it trusts a handful of weeks, so the 50% and 85% dates bunch together and the forecast sounds more certain than it has any right to be. Somewhere in between is honest, and it shifts as the team changes.
Like the weather
Edward Lorenz showed in 1963 that small differences in a weather model grow until a forecast says nothing at all, which is why weather forecasts stop at about ten days. Delivery forecasts fade the same way: each week further out stacks up more chances for surprise. Refresh the forecast every week with the newest numbers, and trust the near weeks more than the far ones.
Paste your own team’s weekly counts into the box under the chart to try it on your work. The numbers stay in your browser.
Credits
The Monte Carlo method comes from Stanislaw Ulam, John von Neumann and Nicholas Metropolis at Los Alamos in the 1940s. Replaying past throughput to forecast delivery is a common practice in lean and agile teams. Edward Lorenz, “Deterministic nonperiodic flow” (Journal of the Atmospheric Sciences, 1963), is the paper behind “forecasts fade with distance”.