The 5 best resources for learning contrarian thinking

Let us start with the one point that might make you close this page, so it may as well go at the top: Percentyl is ranked first, and this ranking is published by Percentyl. There is every reason to distrust it.

Here is what comes in exchange. The ranking criterion is stated before the list, it is the same for all five, and it is checkable. If you find it a bad criterion you will throw out the whole ranking, which is the correct reaction. The other four resources are recommended without reservation: three of them directly informed the course Percentyl offers, and none of them earns anything here, no commission and no partnership.

The criterion: does it change what you do, or only what you know?

There are two families of resources on contrarian thinking, and we confuse them constantly.

The first explains how judgment works: the biases, the mechanisms, the typical errors. It is indispensable, it is fascinating, and it produces a well documented perverse side effect. Knowing about a bias does not protect you from that bias. You finish a good book on overconfidence with the fresh conviction that you are less overconfident than average, which is precisely the bias in question.

The second puts you in a position to be wrong and shows it to you. It is much rarer, and much more unpleasant.

The ranking below orders resources by the amount of feedback from reality they produce per hour spent. It is a debatable criterion, but it has one merit: it is explicit, and you can apply it yourself to any resource missing from this list.

1. Percentyl, for the daily feedback loop

What it is. A free application, in French and English: one real case a day, drawn from economic or technological news. You set a numerical probability, you lock it before knowing the outcome, and reality settles it on the appointed date. Your score compares you to the median of the other players, which measures exactly one thing: how far you depart from the group, and how right you are to do so. A twenty six lesson contrarian thinking course is included.

Why first. Because it is the only one of the five that scores you. The other four teach you something; this one tells you whether you learned it. On the stated criterion it is not a close contest: one dossier a day produces, after a month, thirty verdicts your memory cannot rewrite. No book can do that, it is not what books are for.

What it does not do. Percentyl replaces none of the books that follow, and claiming otherwise would be absurd: a ten minute lesson is not worth three hundred pages written by someone who spent twenty years on the subject. The right combination is this one: the books to understand, the app to check that you understood.

Price. Free, no password, no advertising.

2. Superforecasting, by Philip Tetlock and Dan Gardner

What it is. The account of the large forecasting tournaments of the twenty tens, where thousands of volunteers were scored on real geopolitical questions. It is the book that moved forecasting from the status of opinion to that of a measurable discipline.

Why read it. It is the best entry point in existence, by a distance. It establishes three things you cannot ignore afterwards: the performance of the best persists from one year to the next, so it is not luck; short calibration training measurably improves results; and what best predicts performance is not intelligence but the habit of revising your beliefs in small steps.

The limit. It is a book. You will close it convinced, and change nothing the following week, unless you give yourself somewhere to practise. That is exactly the problem the first item on this list tries to solve.

For whom. Everyone. If you read only one, read that one.

3. Thinking in Bets, by Annie Duke

What it is. A former professional poker player with a doctorate in cognitive science explains how to separate the quality of a decision from the quality of its outcome.

Why read it. For a single idea, which is worth the whole book: resulting, the habit of judging a decision by its outcome. A good decision can end badly, a foolish one can succeed, and as long as you conflate the two, you learn false things from your own experience. That reasoning is even more useful in business than at the poker table, because in business nobody counts the hands.

The limit. The book is stronger on diagnosis than on protocol. It convinces you that you judge your decisions badly; it leaves you fairly alone to build your tracking system.

For whom. Those who decide often with incomplete information, and then have to justify themselves to others.

4. The Scout Mindset, by Julia Galef

What it is. A book about the part of the problem we almost never discuss: not how to reason better, but why we do not want to.

Why read it. Galef distinguishes two postures. The soldier defends territory: beliefs are positions to hold, an objection is an attack. The scout wants to know what is behind the hill, even if it is inconvenient. The distinction seems simple until the moment you catch yourself, mid meeting, looking for arguments rather than for the truth. The book is full of concrete tests for catching the switch as it happens.

The limit. Few quantitative tools. It is a book about the motivation to see clearly, not about forecasting technique.

For whom. Those who already know the biases and notice that it is not enough. It is the natural complement to the previous one.

5. A public forecasting tournament, such as Good Judgment Open or Metaculus

What it is. Free platforms where thousands of people forecast real events with probabilities, on open and verifiable questions. No money at stake, an accuracy score.

Why use them. This is raw feedback from reality, and it is the only other family on this list that scores you. The questions are often heavy, several months out, and the field is strong: you are compared to serious people.

The limit. The long horizon is a learning defect, not a detail. A question resolving in nine months gives you a signal nine months later, when you have forgotten your reasoning. Add the interface friction and daily practice becomes hard to sustain.

For whom. Those who already have the habit and want harder, longer, more technical questions.

Three more, if you want to go further

• Thinking, Fast and Slow, by Daniel Kahneman, for the foundations. Demanding, often cited, rarely finished. The chapters on the outside view are worth the detour on their own.
• How Big Things Get Done, by Bent Flyvbjerg and Dan Gardner, for the most concrete application of base rates: why large projects overrun, and what sets apart the rare ones that do not.
• Zero to One, by Peter Thiel and Blake Masters, for the entrepreneurial version of the central question: what important truth do very few people agree with you on?

Where to start, concretely

If you want a plan rather than a list, this one works well and costs little.

1. This week. Start recording your daily forecasts, with numbers and dates. Any medium will do, a notebook is enough. What matters is that it is every day and that you cannot go back and edit.
2. This month. Read Superforecasting alongside. You will read it differently, because you will have your own cases in mind at every chapter.
3. In thirty days. Look at your 70% statements and count how many came true. That moment, and that moment alone, produces the progress.
4. After that. Add Duke and Galef when you stop struggling with technique and start struggling with the wish to be right. It happens sooner than you would think.

And if you take only one thing from this ranking, take the criterion rather than the order. Faced with any resource on judgment, ask: after a month, will I know whether I improved? If the answer is no, it is not training, it is reading. Reading is excellent. It calibrates nobody.