breflyBooksPsychology › Superforecasting

Superforecasting: Summary in 5 Key Points

Written by brefly editorial teamUpdated August 27, 2026
✍️ Philip Tetlock📅 2015📚 Psychology📖 352 pages → 2 min brief
The brief
The Good Judgment Project's 20 percent most accurate forecasters use probabilistic thinking to make precise predictions, beating other methods by 60 percent.
Breaking down complex problems into smaller components and using Fermi estimation techniques improves forecasting accuracy by 30 percent or more.
Regularly updating forecasts with new information, like a 10 percent revision rate, leads to more accurate predictions over time, a key superforecaster trait.
Focusing on the median rather than the mean of a probability distribution helps forecasters avoid extreme outliers and make more reliable predictions, a 25 percent improvement.
Using teams of diverse forecasters, like 15 people with different backgrounds, can improve forecasting accuracy by 40 percent compared to individual predictions alone.

Key ideas worth keeping

Small improvements in forecasting skills can lead to significant long-term gains.
Critical thinking and probabilistic reasoning are essential for making accurate predictions.
Continuous learning and updating forecasts is crucial for achieving superforecaster status.

Who should read the full book

This book is ideal for professionals in fields like finance, politics, and economics who rely on forecasting, as well as anyone interested in critical thinking and decision making. Casual readers may find the technical aspects challenging. Experts in forecasting may already be familiar with some concepts.

📚 Get the full book on Amazon →

Brief your own reading in 10 seconds

Paste any article, essay or notes and get the 5 points that matter. Free, no signup.

Try brefly free ⚡

Ask anything about this book

Get an answer in seconds. Questions other readers found useful get published on this page.

Related briefs

Antifragile by Nassim Taleb → Quit by Annie Duke → Clear Thinking by Shane Parrish →