Reference class forecasting
Predicting a project's outcome from the track record of similar past projects rather than from its own details.
What it means
Reference class forecasting estimates a future outcome by identifying a class of comparable past cases, examining their distribution of outcomes, and positioning the current case within it — deliberately substituting the outside view for the inside view. It directly counters the planning fallacy and optimism bias, which arise when people forecast from the specifics and best-case narrative of their own project while ignoring how similar efforts actually turned out. Developed from Kahneman and Tversky's work and operationalized by Flyvbjerg for large infrastructure, it has been adopted by some governments to de-bias cost and schedule estimates. Its difficulty is choosing the right reference class and obtaining honest data on past cases. It matters because the base rate of comparable projects is usually a far better predictor than any team's confident bottom-up plan.
Examples
Instead of trusting a rail project's in-house budget, planners look at the cost overruns of 200 similar rail projects and adjust the estimate sharply upward.
Rather than trust the builder's six-week estimate, a homeowner asks four friends how long their kitchen refits actually took and budgets for the twelve weeks the answers suggest.
A student ignores her own optimistic dissertation timetable and asks how long last year's cohort really took, then plans from that spread rather than from her best case.
First described in Kahneman & Tversky (1979); Bent Flyvbjerg.