Confirmation bias

Here is a strange and unsettling fact about studying a company you already like: the more pages you read, the less fair you can become. Most skills improve with effort. This one can rot with it. A reader can spend a whole weekend on an annual report, a dozen broker notes and forty posts, finish exhausted and certain — and be more wrong at the end than at the start, because every hour was spent gathering reasons to believe something already decided.

Picture it plainly. You buy a stock, or you simply form a warm view of one. From that moment a quiet switch flips inside you. Good news about it now feels central, obvious, confirming. Bad news starts to feel temporary — a one-off, a misunderstanding, something the market has already priced in. You are still reading. You are still, by any outward sign, doing research. But the job the reading is secretly performing has changed from finding out whether you are right to proving that you are.

That switch is confirmation bias, and of all the reflexes on this shelf it may be the most expensive, precisely because it wears the costume of diligence. It does not feel like a bias. It feels like being thorough. This module is about catching it in the act — and about the single habit that turns your reading back into an honest instrument.

Confirmation bias does not arrive at the start. It arrives after attachment. Before you own a stock — or a strong opinion about it — your mind is reasonably even-handed; evidence for and against gets weighed at roughly the same rate. The moment you take a position, emotional or financial, the accounting changes. Now you have something to defend, and the mind, which hates the discomfort of having been wrong, begins to sort incoming facts not by how true they are but by how useful they are to the belief you already hold.

Psychologists call the engine underneath this motivated reasoning: reasoning aimed at a conclusion you already want, so that the mind recruits arguments for the destination instead of testing whether the destination is right. It is not lying — that is what makes it dangerous. You genuinely believe you are being fair. The excuse for the rising receivables feels like insight, not evasion. The dismissal of the promoter selling feels like sophistication, not denial.

This is why the module sits where it does, straight after anchoring. An anchor — a price you paid, a target you fixed, a view you formed — is the thing you become attached to; confirmation bias is the machinery that then defends the anchor against every fact that would dislodge it. Anchoring plants the flag. Confirmation bias guards it.

One reassurance before the mechanics, because tone matters here. Noticing this in yourself is not a sign that you are gullible or biased more than others. Every human runs this program; the sharpest analysts run it hardest, because they are best at generating clever reasons for what they already believe. Intelligence is not a defence — it is an accelerant. The only real defence is a method, and the method is refreshingly simple once you see it.

The bias runs in two moves, and naming them is most of the cure.

The first move is selective attention. Facts that match your view are noticed instantly, filed as important, and remembered. Facts that contradict it are noticed slowly, treated as a nuisance to be resolved, and quietly forgotten. Read the same annual report holding two different views and you will genuinely see two different documents — the same words, but a different set of them lit up.

The second move is the deeper one, and it is the heart of this module: asymmetric standards of proof. Without noticing, you run two courts inside your head. Good news is tried in a friendly court: one witness, a nod, admitted. Bad news is tried in a hostile one: it must clear a high bar, survive cross-examination, prove itself beyond doubt — and if it cannot, it is dismissed. Good news needs little proof; bad news faces a committee. The two standards feel like fairness because each verdict, on its own, seems reasonable. It is only when you set them side by side that the double standard shows.

The damage is that this asymmetry is invisible from the inside. You never catch yourself demanding more proof of the bad news, because at each moment you are simply "weighing the evidence." The gap only appears when someone — or a written rule — forces the two standards to become one.

Set the abstract mechanism into an ordinary investing life and it shows up in three habits, each of which feels harmless and each of which costs money.

You curate your reading to agree with you. After buying, you drift toward the bullish posts, the optimistic threads, the accounts whose take you already share — and you follow more of them. The feed slowly becomes a hall of mirrors. It feels like staying informed; it is confirmation bias built into your information supply before a single fact is weighed. If you have read many bull cases on your holding and not one serious bear case, you have not researched it. You have decorated a decision.

You explain away the red flags one at a time. Rising receivables become "timing." Promoter selling becomes "personal reasons." A cash flow that keeps lagging reported profit becomes "conservative accounting." Each excuse is individually plausible — that is what makes the habit so durable. The problem is never one excuse; it is that you have an excuse ready for everything, so no fact is ever allowed to count against you. A thesis that cannot be dented by any conceivable bad news is not a strong thesis. It is an unfalsifiable one, which is a different and far weaker thing.

You let the goalposts drift. The first weak quarter is a one-off. The second is "still early." The third is "the market not understanding the long-term story." At no point do you decide to lower the standard — it lowers itself, one reasonable-sounding step at a time, until a thesis that should have broken two quarters ago is quietly still alive. Because you never wrote down what would break it, there is nothing fixed for the drifting to be measured against.

Watch the two courts run on a single holding. illustrative

A reader owns an API manufacturer — a maker of the active pharmaceutical ingredients that go into finished drugs — and loves it, because revenue is growing fast, up 28% on the year. Then three quieter facts arrive. Receivables have grown 55%, far faster than sales. The top customer now accounts for 42% of revenue, up from last year. And operating cash flow has trailed reported profit for three straight quarters.

Read through the owner's eyes, the 28% is the whole story and the rest is background. The receivables? "Big new clients pay slower — it's timing." The concentration? "A large, loyal buyer — a sign of trust." The cash gap? "Conservative accounting; it'll catch up." Each answer is plausible. Together they form a thesis that no bad news can touch — which should itself be the warning.

The honest read is not "sell" and not "the growth is fake." It is quieter: the cheering fact and the worrying facts deserve the same standard of proof. Growth that converts to cash is one kind of business; growth that piles up as receivables owed by one customer is another kind wearing the same 28%. Which one you own is decided by the cash arriving, not by how good the story feels. The repair is to write the bear case first, in the form a serious opponent would recognise — the growth is funded by loose credit to one buyer and may never convert — and then name, in advance, the single fact that would settle it: does operating cash flow catch up over the next few quarters, or not?

Now run the double standard yourself. The toy below is that same holding — four cheering facts, four red flags. Read it the way the owning mind actually does, and every red flag gets an excuse and vanishes. Then flip to a single standard and watch the flags you were about to wave through stand back up.

The repair for confirmation bias is not to try harder to be objective. You cannot feel your way to fairness while a view you own is quietly rigging the courts. The repair is structural, and it has three moving parts, all of them writing.

First, invert the search. The natural instinct, once you like an idea, is to gather more support for it. Do the opposite: go looking for the single fact that would prove you wrong — the piece of evidence you least want to find. This is disconfirmation, and it is the whole game. A thesis you have genuinely tried to break, and which survived, is worth a hundred you have only tried to support. Before you read one more bullish note, write the strongest bear case you honestly can — the version a smart opponent who wants your money would make.

Second, write "what would change my mind" before you act. One sentence, set down while you are calm and before the position can start defending itself: the thesis breaks if operating cash flow lags profit for three more quarters, or if the top customer's share crosses 50%. This is the fixed post the drifting goalposts can be measured against. Without it, every bad quarter is negotiable; with it, the third weak quarter is not an opinion to argue with but a line you already drew.

Third, hold both sides to one standard. Whatever burden of proof you demand of the bad news, demand exactly that much of the good. If "management sounds confident" is enough to keep you in, then "receivables are stretching" should be enough to worry you — they are the same weight of evidence. The moment you catch yourself accepting a cheering fact on less proof than you'd demand of a worrying one, you have found the bias red-handed.

The defence is powerful, and like every tool on this shelf it has edges worth naming honestly.

A bear case is not automatically right. Searching for disconfirming evidence is a discipline, not a verdict; the point is to give the opposing view a fair hearing, not to assume it wins. A reader who flips from "everything is fine" to "everything is a red flag" has not escaped the bias — only inverted it.

Some negative evidence really is temporary. Receivables sometimes do rise for honest reasons; a promoter sometimes does sell for a genuine personal need. The skill is not to treat every worrying fact as fatal — that is its own paralysis — but to hold it to the same standard as the good news and let the evidence, not the mood, decide.

And "that's just confirmation bias" can itself become a weapon. Waved at a colleague's inconvenient point, or at genuinely new evidence, the label stops being self-awareness and becomes a way of dismissing what you don't want to hear — which is confirmation bias in a lab coat. The test is always the same: not whose view it flatters, but whether it would clear the bar you set for everything else.

The same handful of moves catch owner after owner. Named once, they are far easier to catch in yourself.

1. Reading only what agrees with you. A feed curated to cheer your holding feels like staying informed and is the bias built into your supply of facts. Read the best bear case before the next bull note, not after.

2. Calling every red flag a one-off. One excuse is fine; an excuse ready for everything means no fact can ever count against you. A thesis nothing can dent is not strong — it is unfalsifiable.

3. Treating management's explanation as verification. The company's own account of its own numbers is the source with the most reason to reassure you. It is a claim to be checked, not a check.

4. Moving the goalposts after each weak quarter. If you never wrote down what would break the thesis, it can survive anything — because there is nothing fixed for the drift to be measured against.

5. Confusing effort with fairness. "I read the whole report" says nothing about whether good news and bad news cleared the same bar. Thoroughness is not the same as honesty.