Resources for calibration & forecasting

This page exists to help you get better at probabilistic thinking — regardless of which tools you end up using, including none of the ones listed here. Every resource below is recommended on its own merits, for what it genuinely does well.

Calibration practice takes different forms: instant quizzes for fast feedback, community platforms for social forecasting, quiet personal journals for private reflection, and foundational reading for understanding what separates skilled forecasters from the rest. These uses are not in competition — many people benefit from combining several of them.

This is a curated guide, not a ranking. Every entry here was checked against the tool's official source before publication; tools change, so verify the current details on each site directly.

Interactive tools & quizzes

These tools give you direct, structured practice — either a quick snapshot of how well your confidence matches reality, or an ongoing platform for making and tracking real predictions.

LessWrong calibration quiz quiz

A quick self-test: you answer trivia-style questions with a stated confidence level, then immediately see how well your confidence matched reality. Useful as a standalone exercise or as a warm-up before committing to longer-term tracking — it is complementary to tools that track predictions over time, not a replacement for either. The LessWrong community has produced and curated several versions of this format; the tag page links to current resources. Fact-check note: verify the best current quiz URL before publishing.
Metaculus community forecasting

A serious community forecasting platform where forecasters make public predictions on questions across science, geopolitics, technology, and more. Its strength is in aggregated, public, social forecasting: you can see how a community of forecasters is leaning, track your performance against others, and learn from high-calibration forecasters. This is a different purpose from a private prediction journal — Metaculus is for public engagement with meaningful questions.
Fatebook personal & team

A fast, well-designed tool for tracking predictions personally or with a small team. It supports reminders, sharing, and collaborative prediction-making, which makes it genuinely good for use with colleagues or friends who want to compare forecasts. If you want a social layer on top of personal tracking, Fatebook is built for that.
PredictionBook retired · read-only archive

One of the longest-running open-source prediction trackers — but it was retired and made read-only in late 2023, and its maintainers now recommend Fatebook. It's still worth viewing as a read-only archive and as a record of how the community shaped prediction-tracking practice over the years.

Books

Two books stand out as foundational reading — one on the psychology of honest thinking, one on the research behind what actually makes people good at forecasting.

Superforecasting: The Art and Science of Prediction — Philip E. Tetlock & Dan Gardner (2015) book

The most evidence-grounded book on what separates skilled forecasters from the rest, drawing on Tetlock's multi-year, IARPA-funded Good Judgment Project research. It covers concrete habits — breaking down questions, updating on new evidence, tracking your record — rather than generic advice. Essential reading if you want to understand the research behind calibration practice, not just the mechanics.
The Scout Mindset: Why Some People See Things Clearly and Others Don't — Julia Galef (2021) book

Focuses on the psychological disposition behind good calibration: the willingness to update your beliefs when evidence points the other way, rather than defending positions you have already committed to. Practical and readable, aimed at anyone who wants to think more honestly rather than just track numbers. Galef co-founded the Center for Applied Rationality (CFAR), which has run calibration training for years.

Research & further reading

The Good Judgment Project (GJP) research

The research program behind Tetlock's Superforecasting, run as part of the IARPA forecasting tournament series. The project produced a large body of evidence on which forecasting habits and personality traits predict accuracy, and spawned the concept of the "superforecaster." Good Judgment Inc. continues to run public forecasting projects and publish accessible write-ups of their findings — worth following if you want ongoing research rather than just the book.

One option for private tracking

If you want to track predictions privately — without a public profile or leaderboard — Hunch is a free, browser-based prediction journal with no sign-up required. Your data never leaves your device. It calculates your Brier score and flags over- and underconfidence as your log grows. It is one option among the tools on this page, suited to a different use case from Metaculus or Fatebook: quiet, private, long-term tracking.

For a factual side-by-side of Hunch alongside PredictionBook and Fatebook, see Hunch vs PredictionBook vs Fatebook. For background on what calibration and the Brier score actually measure, see What is calibration? The Brier score explained. For the case that tracking privately changes what you predict and how honestly, see Why track predictions privately?

Try Hunch — no sign-up →