Edition 1·Updated 2026-08-19·20 min read
Sources
About this book
An overview of the book, with source information for further reading.
Most expert predictions are no better than random chance, but forecasting is not a mysterious gift—it's a skill that can be taught and learned.
Drawing on decades of research and the results of a massive government-funded forecasting tournament, *Superforecasting* reveals that a small group of ordinary people, or 'superforecasters,' consistently outperform seasoned intelligence analysts and acclaimed pundits. This book deconstructs what makes them so good. It isn't their IQ or access to special information; it's *how* they think. By embracing uncertainty, thinking in probabilities, constantly updating their beliefs, and practicing active open-mindedness, these individuals achieve demonstrable, measurable foresight. This book is a practical guide to mastering their methods, empowering anyone to improve their ability to see the future and make better decisions in business, finance, policy, and everyday life.
Movement I
Orient
What this book claims, and who wrote it.
- — The book's one-line promise
- — The author and where the book stands
An exploration of the art and science of prediction, revealing how ordinary people can cultivate a specific set of cognitive skills and habits of mind to become extraordinarily good at forecasting future events.
The need-to-know
Forecasting is a measurable skill that can be demonstrably improved through deliberate practice, clear feedback, and targeted training. The best forecasters ('superforecasters') are not defined by who they are (e.g., credentials, intelligence beyond a certain point) but by how they think: they are actively open-minded, intellectually humble, numerate, and self-critical. Key forecasting techniques include breaking down problems (Fermi-izing), synthesizing the 'outside view' (base rates) with the 'inside view' (case-specifics), and updating beliefs in small, frequent, Bayesian increments.
About this guide
The guide describes these challenges, steps and possible outcomes.
Who it is for
You are building a real capability: Superforecasting.
Challenges
- In practice · Forecasting Accuracy erodes when it is left to instinct instead of method.
- For the reader · You were taught the moves piecemeal, never the whole model.
Steps to consider
- 1Master forecasting training.
- 2Master deliberative practice and feedback.
- 3Master collaborative teaming.
Risks
You stay dependent on instinct, and it fails you when the stakes are highest.
Intended result
Forecasting Accuracy becomes something you produce by design, not by luck.
Set the record straight
What this book sets straight
The common beliefs this book pushes back on.
The future is inherently unpredictable, making any attempt at forecasting a fool's errand.
Foresight is a real, measurable skill. While there are limits, people can cultivate the ability to reliably predict events in the short-to-medium term with accuracy far better than chance.
Accurate forecasting is a rare talent possessed by brilliant experts, gurus, or people with access to classified information.
Superior forecasting is the product of a particular way of thinking and working—a process that can be learned and practiced by any intelligent, curious, and determined person.
A forecaster's confidence is a good indicator of their accuracy.
Confidence often merely reflects the coherence of a story someone has constructed, not its truth. The most famous and confident pundits are often the least accurate because their ideological certainty (hedgehog style) impairs judgment.
Movement II
Map
The book's main ideas and how they connect.
Superforecasting's own model — and how it sits inside its field.
- — 10 constructs the book works with
- — Where it agrees with the field, exceeds it, or fills a gap
The constructs
How they connect (12)
- Forecasting Training → influences → Probabilistic Reasoning
- Forecasting Training → influences → Dragonfly-Eye Perspective Taking
- Deliberative Practice and Feedback → influences → Perpetual Beta Orientation
- Deliberative Practice and Feedback → influences → Disciplined Belief Updating
- Collaborative Teaming → influences → Actively Open-Minded Thinking
- Collaborative Teaming → influences → Dragonfly-Eye Perspective Taking
- Perpetual Beta Orientation → influences → Disciplined Belief Updating
- Actively Open-Minded Thinking → predicts → Forecasting Accuracy
- Probabilistic Reasoning → predicts → Forecasting Accuracy
- Dragonfly-Eye Perspective Taking → predicts → Forecasting Accuracy
- Disciplined Belief Updating → predicts → Forecasting Accuracy
- Cognitive Ability and Knowledge → influences → Forecasting Accuracy
Where this book diverges from the corpus
- fills gap Forecasting Training — fills gap
- fills gap Collaborative Teaming — fills gap
- fills gap Perpetual Beta Orientation — fills gap
- fills gap Actively Open-Minded Thinking — fills gap
- fills gap Cognitive Ability and Knowledge — fills gap
- fills gap Forecasting Accuracy — fills gap
Movement III
Master
Explore the book's ideas, examples and practical tools.
How the author makes the case — section by section, with the book's own tools.
- — 10 sections, in the book's order
- — The book's frameworks, checklists, and worked cases
strong · 1 source
- Superforecasting
This section shows you how to convert vague verbal hedges into numerical estimates and update them coherently as evidence arrives.
Probabilistic Reasoning
The skill of thinking about uncertainty in granular, numerical terms, moving beyond a simple 'yes/no/maybe' framework to distinguish between fine-grained degrees of likelihood. This includes numeracy and the ability to update probabilities according to Bayesian principles.
Why it matters. Collapsing every uncertainty into 'probably' or '50-50' throws away the resolution that separates superforecasters from coin-flippers, and it hides your errors from correction.
Myth
That fine-grained probabilities like 73% versus 70% are false precision that adds nothing over rounder numbers.
Reality
Tetlock's data show the best forecasters use and benefit from granularity down to single percentage points; rounding their 73% to 70% measurably degrades accuracy because that last increment encodes real diagnostic information they extracted from the evidence.
- Use the full 0-100 scale; rounding to 'nice' numbers discards accuracy you already possess.
- Every update should be sized by the likelihood ratio of the new evidence, not by how vivid or recent it is.
- Numeracy is trainable — drilling estimation and Bayesian arithmetic raises your ceiling.
The deep drill-down: 8 operational steps, a worked example from the source, 5 decision rules, 5 failure modes, and the “Scorable Forecast Card” tool. Included when you buy this guide.
The rest of this section
the full written section · 3 how-to steps · 2 pitfalls to watch for. Included when you buy this guide.
Grounded in: Superforecasting
strong · 1 source
- Superforecasting
This section teaches you to build a composite judgment by deliberately gathering conflicting viewpoints and reconciling the statistical outside view with the specific inside view.
Dragonfly-Eye Perspective Taking
The cognitive process of actively seeking out and synthesizing information and judgments from multiple sources and viewpoints—including balancing the 'outside view' (base rates) with the 'inside view' (case-specific details)—into a single, more robust judgment.
Why it matters. Anchoring on a single vivid narrative or your own first framing produces confident forecasts that miss because they never absorbed the perspectives that would have corrected them.
Myth
That perspective-taking means finding the one expert or model that is right and adopting their view wholesale.
Reality
The gain comes from synthesizing many partial and imperfect views into an aggregate, not from picking a winner; the metaphor is the dragonfly's compound eye, where thousands of imperfect lenses combine into superior sight.
- Superforecasting MethodProcess — To produce the most accurate possible forecast by systematically gathering and synthesizing evidence and perspectives.
- Establish the base rate before the story; the outside view is your prior, the inside view is your adjustment.
- Seek views that disagree with each other, not sources that agree with you.
- A synthesized aggregate of imperfect estimates reliably beats any single estimate, including your favorite one.
The deep drill-down: 8 operational steps, a worked example from the source, 5 decision rules, 5 failure modes, and the “Dragonfly Synthesis Worksheet” tool. Included when you buy this guide.
The rest of this section
the full written section · 3 how-to steps · 2 pitfalls to watch for · 1 tool for this section. Included when you buy this guide.
Grounded in: Superforecasting
strong · 1 source
- Superforecasting
This section covers how often and how much to revise a forecast so you stay responsive to real signal without chasing noise.
Disciplined Belief Updating
The practice of revising forecasts frequently but in proportion to the diagnostic value of new evidence, avoiding both under-reaction due to belief perseverance and over-reaction to noise. It represents the behavioral manifestation of Bayesian reasoning.
Why it matters. Updating too little leaves you defending a stale forecast against reality, while updating too much makes you a weathervane whose final estimate reflects the last headline rather than the accumulated evidence.
Myth
That the disciplined forecaster is one who commits to a view and holds firm against the temptation to keep changing it.
Reality
Superforecasters update far more frequently than average forecasters, but in small increments sized to the evidence; the discipline is in the calibration of each revision, not in the refusal to revise.
- Frequent small updates beat rare large ones; the best forecasters revise often.
- The size of an update should track the diagnosticity of the evidence, not its emotional salience.
- Treat your prior forecast as evidence too — don't discard the reasoning that produced it just because something new arrived.
The deep drill-down: 6 operational steps, a worked example from the source, 5 decision rules, 5 failure modes, and the “Diagnostic Update Log” tool. Included when you buy this guide.
The rest of this section
the full written section · 3 how-to steps · 2 pitfalls to watch for. Included when you buy this guide.
Grounded in: Superforecasting
strong · 1 source
- Superforecasting
This section clarifies how much raw intelligence and world knowledge you actually need, and where the returns to them stop.
Cognitive Ability and Knowledge
An individual's baseline level of fluid intelligence, numeracy, and crystallized knowledge about the world. The book argues this serves as a necessary but not sufficient precondition for high-level performance, with diminishing returns beyond a certain threshold.
Why it matters. Overrating this construct leads organizations to recruit for IQ and credentials while neglecting the trainable habits that actually separate top forecasters from merely smart ones.
Myth
That the smartest, most knowledgeable people are automatically the best forecasters.
Reality
Above a modest threshold of intelligence and numeracy, additional raw ability yields sharply diminishing returns; what predicts accuracy is what you do with your mind, not how much of it you have.
- Intelligence is a necessary floor, not a differentiator among capable people.
- Beyond the threshold, technique beats raw horsepower — recruit and train for habits, not IQ points.
- Broad, current knowledge across domains matters more than expert depth in one.
The deep drill-down: 7 operational steps, a worked example from the source, 5 decision rules, 5 failure modes, and the “Threshold-and-Reflection Audit” tool. Included when you buy this guide.
The rest of this section
the full written section · 3 how-to steps · 2 pitfalls to watch for. Included when you buy this guide.
Grounded in: Superforecasting
strong · 1 source
- Superforecasting
This section defines how forecasting performance is actually scored and what a good score requires you to get right.
Forecasting Accuracy
The degree to which a forecaster's probabilistic judgments correspond with actual outcomes over time, typically measured by a proper scoring rule like the Brier score, which rewards both calibration (probabilities matching frequencies) and resolution (correctly assigning high probabilities to events that occur and low to those that do not).
Why it matters. Without a proper scoring rule that rewards both calibration and resolution, you cannot tell skill from luck or improve, because you're grading yourself on stories rather than outcomes.
Myth
That being 'right' about a forecast means the predicted outcome happened, so a 90% forecast that fails and a 55% forecast that fails are equally wrong.
Reality
The Brier score judges the whole probability you assigned, not just its direction; a confident forecast that misses is punished far more than a hedged one, which is why good forecasters are measured across many predictions, not one dramatic call.
- A forecast is graded on the probability assigned, not merely on whether the event occurred.
- Good accuracy requires both calibration and resolution — being neither over- nor underconfident, and being willing to be decisive when warranted.
- Skill only becomes visible over a large sample of resolvable, dated questions.
The deep drill-down: 8 operational steps, a worked example from the source, 5 decision rules, 5 failure modes, and the “Scoreable Forecast Card” tool. Included when you buy this guide.
The rest of this section
the full written section · 3 how-to steps · 2 pitfalls to watch for. Included when you buy this guide.
Grounded in: Superforecasting
strong · 1 source
- Superforecasting
This section covers the short-form instructional intervention that raises forecasting accuracy, and why a modest tutorial produces outsized returns. You get the specific content worth teaching and the mechanism behind the gain.
Forecasting Training
Structured instruction on cognitive biases, probabilistic reasoning, and forecasting best practices designed to improve judgment. The book cites a 60-minute tutorial that improved accuracy by about 10%, demonstrating its efficacy as a lever for improvement.
Why it matters. A one-hour intervention that lifts accuracy roughly 10% is the cheapest, highest-leverage move available before you invest in years of practice.
Myth
People assume forecasting skill is an innate gift or an accumulated wisdom that can't be taught in a classroom setting.
Reality
The Good Judgment Project showed a 60-minute tutorial on bias, base rates, and averaging measurably improved Brier scores across ordinary volunteers; the skill is a trainable set of habits, not a talent you're born with.
- A one-hour tutorial on probabilistic reasoning and bias yielded a ~10% accuracy improvement, making training the first lever to pull.
- Effective training targets three concrete moves — reference classes, averaging, and numeric uncertainty — not abstract 'critical thinking.'
- Training seeds the skill but must be reinforced by practice or its effect erodes.
The deep drill-down: 7 operational steps, a worked example from the source, 5 decision rules, 5 failure modes, and the “Pre-Forecast Debiasing Worksheet” tool. Included when you buy this guide.
The rest of this section
the full written section · 3 how-to steps · 2 pitfalls to watch for. Included when you buy this guide.
Grounded in: Superforecasting
strong · 1 source
- Superforecasting
This section explains how repeated forecasting paired with scored, timely feedback converts raw effort into calibrated judgment. You get the conditions that make practice actually improve you rather than merely accumulate.
Deliberative Practice and Feedback
A systematic process of making numerous forecasts, receiving clear, timely, and unambiguous feedback on their accuracy (e.g., Brier scores), and analyzing both successes and failures to refine one's mental models and improve calibration.
Why it matters. Without clear feedback loops you can forecast for decades and never improve, mistaking experience for expertise while your calibration stays flat.
Myth
Practitioners believe that simply making lots of predictions over a long career naturally sharpens their judgment.
Reality
Volume alone teaches nothing; improvement requires quantified scoring (Brier scores), unambiguous resolution dates, and post-mortems on both hits and misses — most professional forecasters never get this and so plateau early.
- Feedback must be scored, timely, and unambiguous — Brier scores on dated questions — or practice produces no learning.
- Analyze successes as rigorously as failures to separate skill from luck.
- Experience without measured feedback creates confidence, not accuracy.
The deep drill-down: 7 operational steps, a worked example from the source, 5 decision rules, 5 failure modes, and the “Forecast Feedback Log” tool. Included when you buy this guide.
The rest of this section
the full written section · 3 how-to steps · 2 pitfalls to watch for. Included when you buy this guide.
Grounded in: Superforecasting
strong · 1 source
- Superforecasting
This section shows you how to structure a forecasting team so it produces judgments sharper than any member could reach alone. You get the design principles that separate a superteam from a committee.
Collaborative Teaming
The practice of working in groups ('superteams') structured to foster constructive confrontation, information sharing, and collective problem-solving, thereby mitigating individual biases and aggregating diverse perspectives into a more accurate collective judgment.
Why it matters. A well-built team can lift forecasting accuracy by double digits over solo forecasters, while a badly-run one merely amplifies the loudest member's error at higher confidence.
Myth
Practitioners assume that putting smart forecasters together automatically averages out their individual errors and yields a better estimate.
Reality
Groups default to conformity, deference, and shared blind spots unless friction is deliberately engineered; accuracy comes not from aggregation but from members actively hunting for information their colleagues missed and challenging each other's reasoning.
- Log each member's forecast independently before discussion so the group starts from genuine diversity, not a single anchor.
- Constructive confrontation is a designed behavior, not a personality trait — build explicit roles and rules that make challenging one another the expected default.
- Judge the team by whether members change their minds in response to evidence, not by how quickly they reach agreement.
The deep drill-down: 7 operational steps, a worked example from the source, 5 decision rules, 5 failure modes, and the “Superteam Diversity & Confrontation Setup Sheet” tool. Included when you buy this guide.
The rest of this section
the full written section · 3 how-to steps · 2 pitfalls to watch for. Included when you buy this guide.
Grounded in: Superforecasting
strong · 1 source
- Superforecasting
This section shows you how to treat your forecasting ability as software that never ships a final version — always patched, always revised, never declared done.
Perpetual Beta Orientation
A foundational mindset combining a 'growth mindset' (belief that abilities are malleable) and 'grit' (passionate perseverance), manifesting as a relentless commitment to learning, self-improvement, and adapting one's forecasting process based on feedback.
Why it matters. Forecasters who see their skill as fixed plateau early, while those who keep iterating steadily pull ahead over years of practice.
Myth
That being in 'perpetual beta' means chasing every new method, tool, or framework you encounter.
Reality
Perpetual beta is not novelty-seeking; it is disciplined revision of what you already do, driven by scored feedback rather than boredom or fashion. The best forecasters change slowly and only when evidence tells them to.
- Improvement comes from revising the process behind forecasts, not merely tracking whether they came true.
- Change your method only when scored feedback justifies it, not when a shinier approach appears.
- Treat a single bad forecast as one data point, not a mandate to overhaul everything.
The deep drill-down: 7 operational steps, a worked example from the source, 5 decision rules, 5 failure modes, and the “Perpetual Beta Postmortem Sheet” tool. Included when you buy this guide.
The rest of this section
the full written section · 3 how-to steps · 2 pitfalls to watch for. Included when you buy this guide.
Grounded in: Superforecasting
strong · 1 source
- Superforecasting
This section teaches you to hold your beliefs loosely enough to test them, actively hunting for the evidence that would prove you wrong.
Actively Open-Minded Thinking
A cognitive style characterized by the willingness to seek out and seriously consider evidence and perspectives that contradict one's existing beliefs, treating beliefs as testable hypotheses rather than cherished possessions. This counteracts belief perseverance and confirmation bias.
Why it matters. Failing to seek disconfirming evidence is the single most reliable way to be confidently, systematically wrong about the future.
Myth
That open-mindedness means giving every viewpoint equal weight and refusing to commit to a position.
Reality
Active open-mindedness is asymmetric: you commit to a probability, then work hardest to find what could break it. Balance is a symptom of good process, not the goal — the goal is calibrated conviction that survives scrutiny.
- Treat each belief as a hypothesis whose disconfirming evidence you are obligated to search for.
- Open-mindedness is asymmetric effort against your own conclusion, not neutrality between all views.
- Weigh contrary evidence by its quality and source independence, not by how much it stings.
The deep drill-down: 7 operational steps, a worked example from the source, 5 decision rules, 5 failure modes, and the “Belief-as-Hypothesis Worksheet” tool. Included when you buy this guide.
The rest of this section
the full written section · 3 how-to steps · 2 pitfalls to watch for. Included when you buy this guide.
Grounded in: Superforecasting
The playbook — the whole process
Beneath the model sits the practical spine — the end-to-end process the source books lay out. Here it is, in sequence, each broken into the steps you actually run.
The sequence — high level first
Illumination of the parts
Process 1 · named in the source
Superforecasting Method
To produce the most accurate possible forecast by systematically gathering and synthesizing evidence and perspectives.
- 1
Unpack the question into its key components and assumptions.
- 2
Adopt the outside view by identifying a relevant comparison class and determining the base rate for outcomes.
- 3
Adopt the inside view by researching the case-specific details and adjusting the base rate up or down.
- 4
Explore other perspectives by consulting other forecasters, experts, and sources to identify clashing causal forces ('dragonfly eye').
- 5
Synthesize all views into a single judgment, expressed as a precise numerical probability.
- 6
Continuously update the forecast in response to new information, typically in small increments.
- 7
Conduct a postmortem after the outcome is known to identify reasoning errors and learn from both successes and failures.
What's underneath
What the field takes for granted
Every field runs on assumptions it rarely says out loud — the beliefs its advice quietly depends on. We surface the load-bearing ones, where they hide, and when they break. Most guides never tell you this.
Placing the idea
How it compares — and where else it applies
We don't just explain the idea in isolation. We place it: against the alternative it replaces, and beyond the domain it was born in. That's the difference between knowing a method and knowing when to reach for it.
How it compares
vs Traditional Punditry (e.g., Tom Friedman)
Both attempt to make sense of complex global events and forecast future developments.
Pundits rely on storytelling, untested intuitions, and often use vague, untestable language. Superforecasters use systematic, evidence-based methods, express judgments in precise probabilities, and are rigorously scored on accuracy.
This book advocates for an evidence-based revolution in forecasting that replaces subjective punditry with a scientific process of measurement and feedback, proving that a different, superior approach is possible.
vs Nassim Taleb's Black Swan Theory
Both agree that the future is highly uncertain and that overconfidence is dangerous. Both are skeptical of long-range, deterministic predictions.
Taleb argues that what can be forecast doesn't matter, as history is driven by unforeseeable Black Swans. This book argues that many consequential short-to-medium-term events ('gray swans') are forecastable, and that improving this skill is highly valuable. It also posits that the consequences of Black Swans unfold over time in ways that can be forecast.
This book offers a pragmatic, optimistic middle ground, acknowledging limits to foresight while demonstrating that measurable improvements are possible within those limits, a concept Taleb's framework largely dismisses.
Where else it applies
The model, taken beyond its home domain
Business and Corporate Strategy
Companies can use forecasting tournaments and superforecasting principles to improve sales forecasts, project timelines, and strategic planning, moving beyond gut feelings to data-driven probabilistic thinking.
Personal Finance and Investing
Individuals can apply principles like balancing outside/inside views and avoiding overreaction to news to make more rational investment decisions and avoid costly behavioral biases.
Public Policy and Government
Governments can use forecasting tournaments as a tool for 'adversarial collaboration,' testing the likely outcomes of competing policies to depolarize debates and make smarter collective choices.
Medicine and Healthcare
Physicians and public health officials can use superforecasting methods to improve diagnoses, predict disease outbreaks, and assess the likely effectiveness of interventions, mirroring the evidence-based revolution Archie Cochrane championed.
Law and National Security
Intelligence agencies and legal professionals can improve assessments of geopolitical risks or the likelihood of a legal outcome by adopting rigorous, scored forecasting methods over purely intuitive judgments.
Movement III · The run-it-now depth
The Playbook
Practical tools and examples from this guide.
Frameworks
The Ten Commandments for Aspiring Superforecasters
A framework summarizing the key principles and practices that contribute to superior forecasting accuracy, derived from the Good Judgment Project's findings.
Start hereBegin by triaging forecast questions to focus your effort on those that are challenging but not impossible.
◆ The full 10-step framework — included with this guide purchase
Checklists
Ten Commandments for Aspiring Superforecasters
◆ All 10 checkpoints — included with this guide purchase
Case studies — including what didn't work
The Iraq WMD Intelligence Failure (2002)
The US intelligence community (IC) assessed with high confidence that Saddam Hussein possessed weapons of mass destruction (WMDs) prior to the 2003 invasion of Iraq.
The IC produced a National Intelligence Estimate stating Iraq 'has' WMDs, a judgment described by the CIA director as a 'slam dunk.' This assessment was a key justification for the invasion.
After the invasion, no WMDs were found. The event was deemed one of the worst intelligence failures in modern history, triggering major reforms and the creation of IARPA.
Bay of Pigs vs. Cuban Missile Crisis
The same group of advisors under President John F. Kennedy made decisions in two major foreign policy crises.
◆ What happened, and the outcome — included with this guide purchase
Archie Cochrane's Cancer Misdiagnosis
Archie Cochrane, a physician and pioneer of evidence-based medicine, was diagnosed with terminal cancer by a renowned specialist.
◆ What happened, and the outcome — included with this guide purchase
Sherman Kent and 'Serious Possibility'
In 1951, intelligence analyst Sherman Kent co-authored a report stating an attack on Yugoslavia was a 'serious possibility.'
◆ What happened, and the outcome — included with this guide purchase
Templates
Sherman Kent's Words of Estimative Probability
To reduce miscommunication by standardizing the numerical meaning of common probabilistic phrases used in intelligence and policy analysis.
◆ The fillable template — included with this guide purchase
Items marked Paid are included when you buy this guide.
Reading well
How the author makes the case
Reading well means seeing how an argument is built, not just what it claims. Here are the moves this author uses to persuade you — the technique, where it shows up, and what it's doing to you. See the machinery, and you read everything more sharply.
metaphor
The core distinction between 'foxes' (who know many little things) and 'hedgehogs' (who know one big thing) is used throughout the book to describe two fundamental cognitive styles.
To create a simple, memorable framework for understanding the complex psychological traits that differentiate good and bad forecasters.
storytelling
The book begins and frequently returns to the story of Bill Flack, an ordinary retiree from Nebraska, to personify the unassuming but powerful nature of a superforecaster.
To make the abstract concepts of forecasting concrete and relatable, showing that superforecasting is a human skill, not a mechanical process.
simplification_of_complexity
The book explains complex statistical concepts like Brier scores and Bayesian updating using simple analogies (like golf scores or adjusting beliefs bit by bit) rather than mathematical formulas.
To make the scientific underpinnings of the research accessible to a general audience without a background in statistics.
pattern_repetition
The core process of 'try, fail, analyze, adjust, try again' (or 'forecast, measure, revise') is repeated in various contexts, from a baby learning to sit up to John Maynard Keynes's investment career.
To reinforce the central theme that improvement comes from a disciplined, iterative process of learning from feedback.
emotional_appeal
The story of Archie Cochrane's cancer misdiagnosis and the high-stakes drama of the Cuban Missile Crisis are used to illustrate the life-and-death consequences of good and bad judgment.
To invest the topic with a sense of urgency and importance, demonstrating that the quality of our forecasting has profound real-world consequences.
Movement IV
Reflect
Consider the book's evidence, limitations and how it compares with other sources.
How good is it — where the book stands against the field, what it leaves open, and when to trust it.
- — Its positioning, its critique, its blind spots
- — When to apply it — and when not
- — The evidence behind the advice
Book profile · in the corpus
This book’s model, against the corpus
4 of 10 constructs align with the field · 6 the book adds · 0 where it diverges · the field spans 34 constructs across 18 books
Shared with the corpus — corroborated by other sources
This book's own emphasis — where it's the authority
The corpus adds — from 17 other sources this book doesn't cover
The full field guide folds in ideas this book doesn’t reach — drawn from 17 further sources and cited in the full model.
This book’s place in the corpus
Why the corpus guide, not just this book
This book is a primary voice in competency adaptability learning v2; the reconciled field-guide adds the 31 constructs and cross-book reconciliation a single book can't provide.
This guide stands on the shoulders of this book and 17 other sources — every claim cited, the disagreements named.
Read the reconciled Competency Adaptability Learning V2 guide →
Read it closely
The book, examined
Not a summary — a close analytical reading: how the book is built, what it leaves open, and where it falls short.
How it’s built — the argument in order
- 01An Optimistic Skeptic — Frames the paradox: most expert forecasts are no better than chance, yet forecasting is a learnable skill—establishing the book's hopeful but empirically grounded stance.
- 02Illusions of Knowledge / Keeping Score — Diagnoses why pundits fail (overconfidence, vague language, no accountability) and argues forecasts must be scored quantitatively—via Brier scores—to be improvable.
- 03Superforecasters / Supersmart? / Superquants? — Identifies who the top performers are and dismantles the assumption that IQ or credentials explain them, pointing instead to numeracy and thinking style.
- 04Supernewsjunkies? (belief updating) — Uses cases like Tim Minto to show that superior forecasting comes from frequent, small Bayesian updates—finding the middle passage between under- and overreaction.
- 05Techniques of forecasting (Fermi-izing, outside/inside view) — Lays out the method: break questions into sub-hypotheses, anchor on base rates (outside view), then investigate case specifics (inside view) purposefully.
- 06Perpetual Beta and Superteams — Argues the strongest predictor is a growth-oriented, self-critical mindset, and that well-managed diverse teams amplify individual accuracy.
What it leaves unsolved
Does forecasting skill on tournament questions transfer to the messy, long-horizon, high-stakes decisions of real policy and business?
The evidence comes from IARPA questions with defined resolution dates; the book itself notes Minto updates 'more typically...to real work as opposed to tournament questions,' hinting the settings differ but not resolving the transfer gap.
How do you forecast genuinely unique or unprecedented events that lack usable base rates?
The method leans heavily on the outside view as a meaningful anchor, but for truly novel events there may be no reference class, and the book does not fully address what to do when the outside view is unavailable.
How should we handle questions that cannot be neatly scored or resolved within a short window?
The entire training and feedback loop depends on 'keeping score,' yet many important questions are ambiguous, distant, or never cleanly resolve—limiting where the demonstrated method applies.
Where it falls short — a fair critique
The book documents techniques (Fermi-izing, outside/inside view, small updates) but the passages give little guidance on how to weight or reconcile conflicting evidence across competing hypotheses beyond 'research pro and con'—the actual synthesis step is underspecified.
Success is measured within a bounded tournament with clear-cut, short-term resolvable questions selected by researchers; whether these are representative of consequential real-world forecasting is asserted more than established.
There is potential survivorship/selection bias: 'superforecasters' are identified by outperforming across seasons, but with large pools some will beat chance by luck; the framing of consistent skill risks overreading noise as trait, though the book only partly guards against this.
The promotional framing (blurbs, 'anyone can learn') overstates generalizability; the demonstrated effect that mindset like 'perpetual beta' is the 'strongest predictor' is correlational and the causal claim that training produces superforecasters is stronger than the tournament design can fully support.
Before you apply it
Using it well
Where the method fits, who it’s for, and the honest case for and against — so you apply it where it works.
Is this book for you?
For you if
- You make consequential predictions in business, finance, or policy
- You're willing to keep score and be proven wrong
- You enjoy methodical, iterative analysis
Skip it if
- You want certainty rather than probabilities
- You won't tolerate constant belief revision
- Your decisions have no measurable outcomes
Need first
- Basic numeracy and comfort with probabilities
- Willingness to embrace uncertainty
- Access to feedback on your predictions
When it applies — and when it doesn’t
- Forecasting questions with clear timeframes and resolvable outcomes — the methods depend on measurable feedback to work
- Repeated, iterative predictions where you can update as evidence arrives — Bayesian incremental updating shines with continuous new information
- Team settings that foster diverse views and constructive criticism — well-managed superteams amplify accuracy
- One-off vague questions with no way to score outcomes — without clear feedback the skill can't be trained or verified
- Truly chaotic, long-horizon events beyond any base rate — probabilistic discipline helps little where uncertainty is irreducible
- Decisions demanding fast commitment without time to research and update — the method is methodical, slow, and demanding by design
The honest case
Strongest case for
- A large government-funded tournament produced measurable, repeatable outperformance
- Success traced to teachable habits, not fixed IQ or credentials
- Specific techniques (Fermi-izing, outside-in view, small Bayesian updates) are concrete and demonstrable
- Perpetual-beta mindset predicts improvement, showing skill is trainable
Strongest objection
- Tournament questions are cleanly resolvable, unlike most real-world strategic uncertainty
- Superforecaster success may partly reflect selection and survivorship over many participants
- Methodical updating is time-intensive and hard to scale to daily decisions
- Short-horizon, scorable questions may not generalize to the messy futures people most want to predict
Tensions — choices to make, not settled answers
The model's constructs are drawn from the source element(s); load-bearing constructs are corroborated across the shelf, emerging ones are flagged.
Ideas covered by the source notes
- Forecasting Training
- Deliberative Practice and Feedback
- Collaborative Teaming
- Perpetual Beta Orientation
- Actively Open-Minded Thinking
- Probabilistic Reasoning
Movement IV · Measure · The evidence
The evidence behind the advice
We don’t just assert — we show the research the ideas rest on: the study, its key finding, what it means for you, and the citation to chase it yourself. Then a curated path to go deeper. Grounded, not hand-waved.
The studies
The empirical backing, with findings and citations — trace any claim to its source.
The accuracy of expert political and economic forecasting.
Expert Political Judgment (EPJ)
The average expert performed about as well as a 'dart-throwing chimpanzee.' However, there were two distinct groups: 'Hedgehogs,' who know 'one big thing' and were highly confident but less accurate, and 'Foxes,' who know 'many little things,' were more pragmatic and self-critical, and demonstrated modest but real foresight.
Good forecasting is not about who you are or what you believe, but about your cognitive style. This suggests forecasting is a skill that can be cultivated.
This is the foundational study of the book, establishing the problem of poor expert forecasting and providing the initial insight into the 'how' of good forecasting.
Philip E. Tetlock, 'Expert Political Judgment: How Good Is It? How Can We Know?' (2005)
Improving geopolitical forecasting through tournaments.
The Good Judgment Project (GJP)
Foresight is a real, measurable skill. A select group of 'superforecasters' emerged who were consistently far more accurate than average. Training, teaming, and deliberative practice all significantly improved accuracy. GJP's top forecasters outperformed university competitors and professional intelligence analysts with access to classified data.
Forecasting is a skill that can be systematically cultivated in any intelligent, determined person. Aggregated judgments from trained amateurs can be a powerful and cost-effective forecasting tool.
This is the central study of the book, demonstrating that superforecasting is real and explaining how it works.
Various papers by Tetlock, Mellers, et al. from 2011-2015.
Test it yourself
Field experiments this shelf implies — designed so you can put the claim to the test.
Hypothesis
Specific interventions, such as cognitive training, organizing forecasters into teams, and creating elite 'superteams,' can significantly improve the accuracy of geopolitical forecasting.
A multi-year, large-scale forecasting tournament (The Good Judgment Project) sponsored by IARPA. Thousands of volunteers were randomly assigned to different experimental conditions: forecasting as individuals, as part of regular teams, or as traders in prediction markets. Performance was tracked and measured using Brier scores.
The primary measure was the Brier score of each individual and team on hundreds of discrete forecasting questions. Secondary measures included psychometric tests for cognitive style (e.g., active open-mindedness), intelligence, and personality.
It was expected that interventions like training and teaming would improve accuracy relative to a control group of untrained individuals, though the magnitude of the effect and the emergence of 'superforecasters' were unknown.
Go deeper
A curated reading ladder — not a dump. Each with why it’s worth your time.
- The Hedgehog and the Fox · Isaiah Berlin
This essay provides the central metaphor used in the book to distinguish between two cognitive styles: 'hedgehogs' who view the world through the lens of one big idea, and 'foxes' who draw on many different ideas and are more adaptable.
- Thinking, Fast and Slow · Daniel Kahneman
Provides the psychological foundation for many of the book's concepts, including the dual-system model of thinking (System 1 vs. System 2), cognitive biases, and the importance of statistical thinking over intuition.
- The Black Swan · Nassim Nicholas Taleb
Presents a strong critique of forecasting, arguing that history is driven by highly improbable, high-impact 'black swan' events that are inherently unpredictable. The book engages with this critique directly.
- The Wisdom of Crowds · James Surowiecki
Popularized the concept of aggregation, showing how the collective judgment of a diverse group can be superior to that of any individual expert. This is a key mechanism for improving forecast accuracy in teams and prediction markets.
- Expert Political Judgment: How Good Is It? How Can We Know? · Philip E. Tetlock
This is the author's prior academic book detailing the 20-year study that established the 'foxes vs. hedgehogs' framework and laid the groundwork for 'Superforecasting'.
Movement V
Measure
Use available assessments to reflect on what you have learned.
A way to gauge yourself against the book's model, and the instruments it gives you.
- — A self-check built from the book
- — Instruments and what you'd measure next
Learning curriculum
After this book, you can…
The book's learning objectives, classified by Bloom's taxonomy and ordered so each builds on the ones before it.
- explainAfter this book you can explain that forecasting is a measurable, learnable skill rather than a mystical gift, citing evidence from the IARPA tournament.Check: Ask learners to summarize the tournament findings and articulate why superforecaster performance disproves the 'unpredictable future' claim.
- distinguishAfter this book you can distinguish skilled forecasters from overconfident experts by identifying the cognitive habits (open-mindedness, humility, numeracy, self-criticism) that account for the difference.Check: Provide profiles of two forecasters and have learners identify which traits predict superior accuracy and why.
- cultivateAfter this book you can cultivate a perpetual-beta mindset—combining growth mindset and grit—and explain why it is the strongest predictor of superforecaster performance.Check: Have learners reflect on and describe habits demonstrating self-improvement commitment and connect them to forecasting performance evidence.
- quantifyAfter this book you can express uncertain judgments in granular numerical probabilities rather than vague verbal terms.Check: Present scenarios and have learners assign calibrated probability estimates and justify their granularity.
- decomposeAfter this book you can decompose a difficult forecasting question into researchable sub-components using Fermi-ization.Check: Give learners a complex prediction question and have them break it into tractable hypotheses and estimable parts.
- calculateAfter this book you can measure forecasting accuracy using a proper scoring rule such as the Brier score.Check: Give learners a set of probabilistic forecasts and outcomes and have them compute and interpret Brier scores.
- seekAfter this book you can practice actively open-minded thinking by seeking out and seriously weighing evidence that contradicts your beliefs.Check: Assign a forecast and require learners to document contrary evidence they gathered and how it altered their view.
- reviseAfter this book you can update your forecasts in small, frequent, evidence-proportional increments, avoiding both under- and over-reaction.Check: Give learners a sequence of incoming evidence and have them adjust probability estimates in proportion to diagnostic value, explaining each change.
- integrateAfter this book you can synthesize the outside view (base rates) with the inside view (case specifics) to form an anchored estimate.Check: Have learners forecast an event by first stating a base-rate anchor, then adjusting with case-specific evidence, documenting each step.
- designAfter this book you can design and run a deliberate-practice cycle of 'forecast, measure, revise' using clear, timely feedback.Check: Have learners set up a personal forecasting log with scored predictions and a revision routine, then evaluate improvement over time.
- synthesizeAfter this book you can synthesize multiple viewpoints and sources into a dragonfly-eye perspective on a problem.Check: Have learners aggregate diverse expert judgments and datasets into a single reasoned forecast, showing how each perspective informed the result.
- applyAfter this book you can apply superforecasting methods to a real decision in business, finance, or policy and defend your reasoning.Check: Have learners produce a full forecast on a real-world question integrating Fermi-ization, outside/inside views, probabilistic estimates, and an updating plan.
- structureAfter this book you can structure a forecasting team to foster information sharing, constructive confrontation, and diverse perspectives.Check: Have learners propose team norms and processes for a 'superteam' and justify how each element amplifies accuracy without groupthink.
- evaluateAfter this book you can evaluate the quality of a given forecast and forecasting process, judging calibration, updating discipline, and reasoning transparency.Check: Provide a documented forecast and have learners critique its use of base rates, probabilistic granularity, and belief updating.
Validated instruments — where the research already has a measure
Active Open-Mindedness (AOM) Scale (reconstructed)
validated“People should take into consideration evidence that goes against their beliefs.”
How to measure it
Turning each idea into a measure
For each construct: how to operationalize it, the observable signals to look for, and how well it holds up.
Assignment to and completion of a standardized training curriculum covering topics such as cognitive biases (e.g., confirmation bias), probabilistic thinking, the use of base rates (outside view), and belief updating principles.
- Completion of training modules
- Performance on post-training knowledge checks
- Observed application of trained techniques in forecasting rationales
Typically measured as a binary variable (trained vs. not trained) in experimental settings.
The number of distinct forecasting questions an individual answers over a defined period, combined with their exposure to and engagement with quantitative accuracy scores and qualitative post-mortems after questions resolve.
- Count of forecasts made per user
- Frequency of viewing performance dashboards
- Authoring or commenting on post-mortem analyses of resolved questions
Can be measured by counting forecasting and feedback-reviewing behaviors logged in a system.
Assignment of an individual forecaster to a team with shared goals and communication channels, as opposed to forecasting individually. The effectiveness can be further operationalized by the frequency and quality of intra-team communication.
- Frequency of posts and comments in team forums
- Content analysis of discussions for evidence of debate and synthesis
- Team members' self-reports of psychological safety and collaboration
Primarily an experimental condition (team vs. individual), but team dynamics can be measured via surveys and behavioral observation.
A composite measure based on an individual's scores on standardized psychometric scales for growth mindset and grit, reflecting their underlying disposition toward effortful improvement and resilience to failure.
- High scores on relevant psychometric tests
- Sustained engagement in forecasting over long periods, despite setbacks
- Explicit statements about learning goals and the importance of practice
Measured with validated psychometric scales from Dweck (mindset) and Duckworth (grit).
An individual's score on a validated psychometric scale designed to measure the tendency to be open to belief change and to seek out dissonant information, such as Jonathan Baron's Actively Open-minded Thinking (AOT) scale.
- High scores on AOT scales
- Seeking out news sources from opposing viewpoints
- In rationales, explicitly discussing and weighing evidence that runs counter to their final judgment
Measured using multi-item Likert-type scales.
The degree of precision used in probability estimates (e.g., use of increments of 1% vs. 5% or 10%) and scores on numeracy tests. It reflects an individual's ability and tendency to quantify uncertainty with high resolution.
- Use of non-rounded probability values (e.g., 27%, 83%)
- High scores on numerical reasoning tests
- Rationales that correctly apply principles of probability
Granularity can be measured by analyzing the distribution of an individual's forecast values. Numeracy is measured with standardized tests.
Evidence from forecasting rationales or commentaries that an individual explicitly referenced a base rate or comparison class ('outside view') and integrated it with case-specific information ('inside view') to arrive at a forecast.
- Explicit mention of base rates or historical frequencies
- Discussion of conflicting arguments or scenarios ('on the one hand... on the other hand')
- Citing and integrating views from teammates or diverse news sources
Can be measured through content analysis of forecasting rationales.
A measure of the frequency and magnitude of an individual's forecast updates for a given question. High discipline is characterized by more frequent and smaller adjustments, as opposed to infrequent, large jumps or no adjustments at all.
- Number of updates per question-day
- Average percentage point change per update
- Timeliness of updates following significant news events
Measured by analyzing time-stamped forecasting data.
Scores on standardized psychometric tests measuring fluid intelligence (e.g., Raven's Progressive Matrices), numeracy, and knowledge of current events and politics. The measure represents a baseline cognitive resource level.
- High scores on IQ-type tests
- High scores on current events quizzes
- Correctly answering questions requiring basic mathematical calculation
Measured with standardized psychometric tests.
An average Brier score calculated across a large number (>100) of diverse, closed-ended forecasting questions with unambiguous outcomes. A lower Brier score indicates higher accuracy.
- Low average Brier score (closer to 0)
- Calibration curve close to the diagonal
- High resolution score
Brier scores range from 0 (perfect accuracy) to 2 (perfectly inaccurate), with 0.5 representing random guessing for binary questions.
Requires a large number of forecasts to be a valid measure of skill rather than luck. · Year-to-year correlations among superforecasters are consistently high (~0.65), indicating strong reliability.
Self-assessment
Where would you like to improve?
This is a structured self-diagnostic built from the model — a mirror for reflection, not a validated psychometric scale. For validated measurement, see the instruments below.
1 = Strongly Disagree · 7 = Strongly Agree
Proposed measures — starter instruments where no validated one was found
Forecasting Training Rigor Index
proposed · not validatedRate each observable statement on the scale below.
- A documented curriculum covering cognitive biases and probability calibration is delivered to all forecasters before they issue live forecasts.
- Training materials are updated at least annually to reflect newly identified reasoning errors or methodological advances.
- Completion of forecasting training modules is logged and verifiable for every participant in the forecasting process.
Scale: 1–7 (Strongly Disagree → Strongly Agree). Average the item ratings. This proposed index has no validated cutoff.
Calibrated Feedback Loop Index
proposed · not validatedRate each observable statement on the scale below.
- Every forecast issued is scored against an objective accuracy metric such as Brier score within a defined follow-up window.
- Forecasters receive individualized accuracy reports on a recurring, scheduled basis rather than only upon request.
- The system tracks and displays historical accuracy trends per forecaster so patterns of improvement or bias are visible over time.
Scale: 1–7 (Strongly Disagree → Strongly Agree). Average the item ratings. This proposed index has no validated cutoff.
Structured Team Deliberation Index
proposed · not validatedRate each observable statement on the scale below.
- Forecasting teams hold scheduled sessions where members are required to challenge each other's reasoning before a group estimate is finalized.
- A documented process exists for aggregating dissenting views and resolving disagreements among team members.
- Team composition and discussion rules are explicitly designed to surface diverse information sources rather than defer to seniority or consensus pressure.
Scale: 1–7 (Strongly Disagree → Strongly Agree). Average the item ratings. This proposed index has no validated cutoff.
Sources
- Superforecasting — Philip E. Tetlock
The cheat sheet
Everything, on one page
One key takeaway from each section, with links to read more.
- Forecasting TrainingA one-hour tutorial on probabilistic reasoning and bias yielded a ~10% accuracy improvement, making training the first lever to pull.
- Deliberative Practice and FeedbackFeedback must be scored, timely, and unambiguous — Brier scores on dated questions — or practice produces no learning.
- Collaborative TeamingLog each member's forecast independently before discussion so the group starts from genuine diversity, not a single anchor.
- Perpetual Beta OrientationImprovement comes from revising the process behind forecasts, not merely tracking whether they came true.
- Actively Open-Minded ThinkingTreat each belief as a hypothesis whose disconfirming evidence you are obligated to search for.
- Probabilistic ReasoningUse the full 0-100 scale; rounding to 'nice' numbers discards accuracy you already possess.
- Dragonfly-Eye Perspective TakingEstablish the base rate before the story; the outside view is your prior, the inside view is your adjustment.
- Disciplined Belief UpdatingFrequent small updates beat rare large ones; the best forecasters revise often.
- Cognitive Ability and KnowledgeIntelligence is a necessary floor, not a differentiator among capable people.
- Forecasting AccuracyA forecast is graded on the probability assigned, not merely on whether the event occurred.
Publication details
Consult the sources listed in this guide for further reading. Edition 1 · Updated 2026-08-19.