The Confidence Cone
Build a simple forecast — revenue, demand, headcount — with a trend knob and an uncertainty knob, and watch its confidence cone widen with every month you project. The median line is never drawn alone: it ships with the band of where the number could reasonably land, so you can see exactly how much of any forecast slide is guesswork dressed as precision. Drag a what-if shock (a lost channel) and the whole cone moves live. Runs entirely in your browser (0 uploads, works offline).
How to use it
- Set the monthly growth slider — the trend the median line follows (it can be negative).
- Set the volatility slider — how much uncertainty each month adds. The cone widens with the square root of time.
- Hit what-if: lose your biggest channel to knock points off growth and watch the median and its cone both shift.
- Read the median at month 24, the 80% range (p10–p90), and the cone width as a percent of the number itself.
What this clears up (the fundamentals)
- A point forecast is a distribution with the error bars deleted — "next year: 180" is one draw from a range. The cone is the range you were never shown.
- Uncertainty compounds with √time — the band widens as the square root of the horizon, so a forecast three months out is tight and one two years out is a fan. Distance is doubt.
- The confidence level is a choice, not a fact — the 80% cone (p10–p90) says "8 times in 10 the truth lands in here." A wider interval is more honest, not more precise.
- The cone can be wider than the number — when volatility is high and the horizon is long, the honest range dwarfs the median. That is the tell that the slide is guesswork, however confident the single line looks.
Where it's used
The forecasting baseline behind every revenue plan, demand forecast, and headcount model — and the reason experienced operators distrust a single confident line on a slide. It crosses into unit-economics and finance-reporting work anywhere a projection is presented. It's a miatz build-lab concept playable; learn to build the live-recomputed confidence band yourself.
FAQ
What is a confidence cone (prediction interval)?
The band around a forecast that shows where the true value is likely to fall, not just the single best guess. It widens over time because uncertainty accumulates the further out you project. An 80% cone means the outcome should land inside it about 8 times in 10.
Why does the cone get wider the further out I forecast?
Because each month's uncertainty stacks on the last. The width grows with the square root of the number of months, so short-horizon forecasts are tight and long-horizon ones fan out — distance from today is distance from certainty.
Isn't a single forecast number more useful than a range?
A single number is easier to put on a slide, but it hides the risk. The median tells you the most likely value; the cone tells you how wrong you could reasonably be. Planning against the median alone is how "confident" forecasts miss.
Is anything uploaded?
No. The forecast and its confidence cone are computed entirely in your browser — nothing is transmitted or stored. Turn off your Wi-Fi and it still works.
Limits
A teaching model: the median follows clean compound growth and the cone is a log-normal fan from a single volatility knob. Real forecasts use fitted trend, seasonality, and error terms estimated from history. The lesson — every point forecast has a cone, and it widens with time — is exactly the real practice.
Related
Part of the Demystify Playgrounds. Explore the rest from the Playgrounds home.
Bookmark this page (Ctrl+D, or ⌘D on Mac) — it works offline the next time you need it.