Forecasting is a skill, not a gift

Most people think being good at guessing the market is like being tall - you either have it or you don't. The confident voice on the TV, the friend who "called" the last big rise, the fund manager who had one great year - we assume they can see the future a little more clearly than the rest of us.

A scientist named Philip Tetlock spent decades testing whether that's true. What he found is a bit of a shock: on average, expert predictions about tricky, real-world events are barely better than flipping a coin - and the more famous the expert, the worse they often did. But there was a bright spot hidden in the same research. A small group of totally ordinary people - no special access, no secrets - made clearly better guesses than everyone else, year after year. They weren't gifted. They had a habit.

Here's the habit, and it's almost silly how simple it is: they kept score. They wrote their guesses down with a number and a date, checked them later, and learned from the misses.

Think of two people at a cricket net. One is the player: every practice she scribbles down how many balls she middled and how many she missed. Because she can see her own numbers, she spots what's going wrong and slowly gets better. The other is the loud commentator: he shouts confident opinions all day but never picks up a bat and never checks a single thing. He can talk forever and never improve one bit - because he never keeps score. The pundits on TV are the loud commentator. The whole trick of this book is to be the player instead.

Now sit with the surprising bit for a moment, because it changes how you should feel about the whole game. If forecasting were pure talent - a gift you either have or don't - then reading a book about it would be pointless, the way reading about being tall won't make you tall. But Tetlock's research says the opposite. The good guessers didn't share a special brain; they shared a practice. And a practice is exactly the kind of thing an ordinary person can copy. That's the hopeful heart of this book: you are not stuck with the forecasting ability you were born with. It behaves far more like cycling or swimming - clumsy at first, then steadily better the more honestly you practise - than like height, which is fixed the day you're born.

Here's the everyday version of why score-keeping is the whole secret. Picture Aayra learning to shoot a basketball into a hoop in the dark, with a blindfold on. She shoots, hears nothing, shoots again, hears nothing. She could practise for ten years like this and never improve one bit, because she never finds out where the ball went. Now take the blindfold off and switch on a light. Suddenly every shot tells her something - too short, too far left, too hard - and within a week she's better. Forecasting without keeping score is shooting in the dark. The number and the date are the light switch. Nothing else in this chapter matters until that light is on, because a guess you never check is a shot you never see land.

Keeping score has two parts that sound the same but aren't: being right a lot, and being calibrated. That second word is the one almost nobody checks, so let's make it easy.

Calibration just means: your sureness matches what really happens. When you say you're "70% sure," the thing should come true about 70 times out of 100 - no more, no less. The best example is a weather forecaster. A good one who says "70% chance of rain" is telling the truth if, across all the days she says that, it rains on about 7 of every 10 of them. If someone shouts "I'm 90% sure!" about everything but is only right 6 times out of 10, they aren't unlucky - they're just too sure of themselves. And that single leak drains more money out of everyday investors than any single bad stock ever could.

The fix isn't to be a genius. It's to write the number down - "60% chance," not "I like it" - and check it later. You simply can't see the gap between how sure you feel and how often you're right until you've got both written down side by side.

There's a second half to a good score that's just as important, and it's the opposite worry. Being calibrated is easy if you cheat: just say "50% chance" about everything. You'd be perfectly honest and perfectly useless, because a shrug helps nobody decide anything. So a good score also rewards being decisive - daring to say 20% and 85% when you genuinely know something, and being right when you do. Imagine a doctor who says "maybe you're ill, maybe you're not" to every single patient. Never wrong, never useful. The doctor you want is the one who says "I'm 90% sure this is just a cold" and "I'm 90% sure this one needs an X-ray" - and is right both times. Good forecasting lives in that combination: sticking your neck out and your neck surviving. Playing it safe at 50% forever is just the loud commentator in a lab coat.

So keep two questions in view at once. When you say a number, does it come true that often (calibration)? And are you brave enough to move away from a lazy 50% when you actually have a view (decisiveness)? The player at the cricket net needs both: she has to take real shots, not gently pat every ball back, and she has to record honestly how many she middled. One without the other teaches you nothing.

Let's try it on your own guesses. illustrative

Say that over one year you make ten confident calls - "this stock will beat the market," "this one has stopped falling," "these results will be a letdown." Each time, you felt about 80% sure. A year later you go back and check: six of the ten actually went your way.

That gap is worth real money. You said 80% but you delivered 60%. If you'd been betting big because of that 80% feeling - say putting ₹80,000 of a ₹1,00,000 pot behind each "sure thing" - you were betting as if you were far more reliable than you truly are. Someone who quietly noticed "my 80% calls really only land 60% of the time" would bet smaller, keep more money spare for the misses, and survive the cold streak that always comes eventually. The person who never keeps score just keeps shoving ₹80,000 onto a bet that's barely better than a coin flip.

And notice what the scoring did not need: no secret information, no cleverer idea. It only needed you to write down the number and the date, then look back. Feeling a bit humbled is the whole reward - it's what makes you bet more sensibly next time.

Now sit with what that gap actually costs, because it's easy to shrug at "80% versus 60%" as if it were a small rounding error. It isn't. A person who truly lands 80% of the time is right four times for every miss; a person who lands 60% is right three times for every two misses. That's nearly double the failure rate - and every extra miss is real rupees leaving your pocket. The frightening part is that both people feel exactly the same amount of sure. The 80% feeling and the 60% reality sit side by side in your head, and you cannot tell them apart from the inside. The only thing that ever pulls them apart is the score-sheet. Without it, the over-sure investor doesn't feel over-sure - he feels like the sharpest person in the room, right up until the cold streak arrives and quietly empties his account.

Here's the deepest reason keeping score matters, and it's the thing that separates a real skill from a lucky streak. In a game full of chance - and the market is soaked in chance - a single result tells you almost nothing about whether the thinking behind it was any good. A careful, well-reasoned call can still lose to bad luck. A reckless, silly guess can still win on good luck. This time. If you judge yourself by the last outcome, you'll learn everything backwards: you'll praise the lucky reckless call and abandon the sound careful one after an unlucky patch.

Think of a class of a hundred students all flipping a coin, guessing heads or tails. After six rounds, pure chance guarantees that one or two of them will have guessed right every single time. Put that lucky student on stage and they look like a coin-guessing genius. Hand them a microphone and they'll happily explain their "method." But there's no method - only a big crowd and a bit of luck picking a winner. The famous fund manager who had one dazzling year is often exactly this student: plucked out of a huge crowd by chance, then mistaken for a wizard. The only way to tell a genius from a lucky flipper is to watch them over many calls and see if the winning keeps up.

This is why the score-sheet has to be long, not just honest. Judge yourself on how good and how honest your calls are across dozens of tries, never on the one win you happen to remember. In a game full of luck, one good decision can end badly and one bad decision can end well - so grade the thinking across many scored calls, not the single result you got. A short record is a coin flip wearing a costume; only a long one starts to tell you the truth about yourself.

Look around at where investing tips actually reach you, and you'll see the loud commentator everywhere - just with an Indian accent. The confident anchor on a business channel calling "levels" for tomorrow. The WhatsApp group where someone forwards "multibagger alert - buy before Monday!" The relative at a wedding who "made a killing" on one stock and never mentions the four that sank. None of them keep score. None of them will show you a dated list of every call they made and how those calls actually turned out. That missing score-sheet is the tell.

So the single most powerful thing you can do is boringly simple, and almost nobody around you does it: keep your own score-sheet. Next time a tip arrives - from the TV, the WhatsApp forward, the wedding uncle, or your own excited brain - before you act, write one line in a notebook or a phone note: today's date, the stock, an actual number ("I'm 65% sure this beats a plain index fund over the next year"), and a deadline. Then let the year pass and check. Do this for a dozen tips and something quietly wonderful happens: you stop needing to trust or distrust anyone. You have data on them, and on yourself. The forwarded "sure things" reveal their true hit-rate, and so do your own hunches. In a market this full of loud, unscored voices, the person quietly keeping a dated notebook has an edge that costs nothing but a little honesty.

The trap is that a confident story feels exactly like a real guess - but you can never mark it right or wrong.

"This is a great company with a long road ahead" can never be proven wrong. No fact that ever happens could contradict it, so it can never make you better. It's comfy because it's safe from ever being checked. That's the loud commentator's whole game: say enough vague, hedged, memorable things, and later you can point at the one that came true and quietly forget the pile that didn't.

Score-keeping is powerful, but like any tool it can be swung the wrong way, so here are the honest cautions.

First, a short score-sheet lies just as loudly as no score-sheet. Ten calls cannot separate skill from luck - as the coin-flippers showed, chance alone will hand a few people a spotless record over a short run. If you keep score for a month, get seven right, and crown yourself a superforecaster, you've simply swapped a story for a slightly-bigger story. Real signal needs many calls across many months, ideally covering a bad market as well as a good one, because anyone looks brilliant when everything is rising. Treat a short record as a rough hint, never a verdict.

Second, you can quietly cheat your own targets. If your written call is vague enough - "this should do well over time" - then every outcome can be squinted into a win, and the score-sheet becomes a comfort blanket instead of a mirror. The whole value comes from targets so specific that a stranger reading them a year later would agree, without argument, whether each one hit or missed. If you find yourself relieved that a call is "still playing out," that's usually a sign the target was too soft to ever be marked.

Third, keeping score is a way of reading, not a way of predicting the unpredictable. Some things genuinely cannot be forecast - a sudden global shock, a freak event, next week's exact price. Scoring won't grant you a crystal ball, and chasing precise short-term predictions is a good way to fill a notebook with noise. What scoring does is humbler and more useful: it slowly shows you which kinds of judgements you're actually decent at and which you should bet small on. And a reminder that outlives this whole book - none of this is a licence to tip stocks, least of all to yourself. The point of a good score is to size your bets honestly and stay humble, never to convince yourself you've found a sure thing.