How Hedge Funds Actually Research Companies

Hedge funds research companies by starting from primary sources instead of headlines, reading the filings and the notes and the earnings call themselves, and then spending most of their effort trying to prove their own idea wrong. The output they want is not a summary of the business. It is a variant perception: a specific, checkable reason their view of the company differs from what the market is already pricing.

That last part is the whole game. Anyone can describe a business. The professional edge comes from finding a place where the crowd is looking at the wrong number, or reading a fact too optimistically, or too harshly, and then holding that view only as long as the evidence supports it. Our subject here is the habits that produce that edge. It is not a step by step checklist. For the linear version of the process, see the institutional equity research workflow and the broader equity research process. Here we are talking about how the good ones think.

They start from the source, not the story

Retail research usually starts with a conclusion someone else reached: a news headline, a broker’s one-liner, a screenshot of a rising chart, a tip. Professional research starts one layer below all of that, with the primary documents the company itself files.

The reason is simple. Every secondary source is already an interpretation, and interpretations lose information. A headline that says “profit up 20%” has thrown away the reason profit was up, which is the only part that tells you whether it will happen again. So the analyst goes to the annual report, the quarterly filing, the notes to the accounts, and the earnings call transcript, and reads them in that order of trust.

The notes to the accounts are where the real work often is. The face of the income statement is designed to be read quickly; the notes are where accounting policies, one-off items, related-party transactions, and segment detail actually live. A single consolidated topline hides a great deal. Take Reliance Industries in the quarter ended September 2025. At the segment level its oil-to-chemicals business did roughly ₹1,60,600 crore of revenue at about a 9% segment margin, while its digital services arm did roughly ₹42,700 crore at about a 52% margin, and the upstream oil and gas segment did only about ₹6,100 crore but at roughly an 83% margin. The biggest revenue segment is one of the thinnest, and the fattest margins sit in much smaller lines. You cannot see any of that from the headline number. You see it only if you go and read the segment disclosure. Learning to do this yourself is the point of how to read an annual report.

They hunt for the reason they are wrong

The habit that most separates professional from amateur research is uncomfortable: you spend your best hours attacking your own idea.

Most people, once they like a stock, go looking for reasons to like it more. This feels like research and is actually the opposite. Confirming evidence is easy to find for almost any story, so finding it tells you nothing. The information that changes your mind is the evidence that cuts against you, and you only get it if you deliberately go looking.

So a good analyst writes down, before anything else, the two or three things that would have to be true for the idea to work, and the specific facts that would prove it broken. Then they go hunting for the second list. If the thesis is “margins will expand because the new plant lowers cost per unit”, the disconfirming questions are concrete: is the plant actually running at capacity, is the cost saving showing up in the segment margin, is a competitor adding supply that will compete the saving away? An idea that survives that treatment is worth something. An idea you were never willing to test is just a hope with a ticker.

The market pays you for being right in a way that others are not yet. It does not pay you for agreeing with it loudly.

They form a variant perception

Out of all that reading and stress-testing, the professional is trying to land on one thing: a view that is both different from consensus and defensible. This is the variant perception, and it is the spine of any real investment memo.

A variant perception is not “this is a great company”. The market may already know that, and the price may already reflect it, in which case being right about quality earns you nothing. The variant perception is a claim about where your read of the facts differs from the market’s read, stated so plainly that you could later be shown wrong. “Consensus is modelling this business as a slow grower and we think the order book says otherwise, and here is the evidence.” That is a position. “I like the management” is not.

Crucially, the variant view has to be falsifiable. If there is no future fact that could disprove it, it is not analysis, it is faith. The discipline of writing the thesis as something checkable is what lets a desk grade itself honestly later, which is the only way anyone actually improves.

They do channel checks

Filings tell you what has already been reported. They do not tell you what is happening in the field right now, and they are written by the party with the most incentive to frame things kindly. So professional desks gather their own primary evidence by talking to the people around a business: distributors, suppliers, customers, former employees, industry consultants, and specialists.

This is what “channel checks” means. If a consumer company says demand is strong, you can call a few distributors and ask whether they are actually being asked to hold more stock. If a software firm says deal sizes are growing, you can ask people who sit across the table in those deals. None of this is exotic. It is just the ordinary work of not taking a self-interested claim at face value. The aim is to see whether the story management tells matches the reality other people are living, before that reality shows up in a printed number two quarters later.

For retail investors, full channel checks are hard, but the principle scales down. Reading the same company’s competitors and suppliers, and reading what customers say, is a version of the same instinct: triangulate the claim against sources that have no reason to flatter it.

They treat monitoring as part of the research

On an amateur desk, research ends when you buy. On a professional one, buying is roughly the halfway point, because the thesis was written as a set of things that should happen over time, and now you have to check whether they do.

This is why the exact words in an earnings call matter, and why serious analysts keep a running record of them. When Asian Paints guided, for its forward outlook, to high single-digit volume growth “in the band of about 8-10%” and pointed to an 18-20% EBITDA margin band while speaking of “maintaining our margin guidance”, that is not a headline to skim. It is a dated, numbered promise with a hedge built in. The professional writes it down and then, quarter after quarter, checks whether volumes and margins are tracking toward that band or drifting away from it, and watches whether the language shifts from confident to careful over time. The drift in tone often arrives before the drift in numbers. This kind of close reading is what we walk through in Indian concall analysis examples.

Monitoring is also where you catch yourself. A new fact that breaks part of your thesis is not an inconvenience to be explained away; it is the most valuable thing you will get all quarter, because it is telling you to change your mind before the price does.

The mindset in one line

Strip away the tools and the jargon and the professional research mindset comes down to a few habits. Go to the source. Look hardest for the evidence that you are wrong. Hold a view only as sharp and as checkable as the facts allow. Test what management says against people who have no reason to flatter it. And never stop watching, because a thesis is a claim about the future, and the future keeps voting.

None of this requires a Bloomberg terminal or a research team. It requires the willingness to read the boring documents, the humility to attack your own idea, and the discipline to write down what would prove you wrong. That is a way of thinking, and it is available to anyone patient enough to practise it.