Analyst reports

A friend forwards you a PDF. It is a research note from a big brokerage, ten pages of charts and tables, and right at the top, in bold: BUY. Target ₹1,240. The stock trades at ₹980 today. Your eye does the arithmetic before your brain catches up — that is about 27% upside — and a small warm feeling arrives: someone who does this for a living has done the work, and the work says yes.

Hold that feeling still for a moment, because this module is about taking it apart. An analyst note is one of the most useful documents a retail investor can get for free, and one of the most misread. The trick is that the two things you naturally look at first — the rating and the price target — are the two things worth the least. The parts worth the most are buried below them, in the workings almost nobody scrolls to.

So the question for this module is narrow and practical: when a brokerage note lands in front of you, which lines do you keep, which do you ignore, and who profits if you swallow the bold bit at the top whole?

Start where this shelf always starts — with the money. The analyst who wrote that note is almost never paid by you. She works on the sell-side: the research desk of a broker. Ask how that broker earns its living and the incentives fall out on their own.

It earns three ways, and none of them is "you make good returns." It earns brokerage — a small cut every time anyone buys or sells, so more trading is more revenue, and a note that makes you do something is worth more than one that makes you sit still. It earns from investment banking — helping companies raise money and do deals — which means the companies the analyst covers are also potential clients the firm would love to keep happy. And it earns from asset management and large clients — the big institutions who get the research first and whose trading commissions pay the desk's bills.

Look at that list and the shape of the problem is obvious. The person the research is really for is the institution that generates commissions, not the individual who downloaded the PDF for free. You are reading someone else's mail. That is not a scandal and it is not a reason to bin it — the mail can be excellent — but it is the reason the tone leans optimistic and the reason the word "Sell" is so strangely absent. Follow the money before you weigh the argument, and the rating starts to look less like a verdict and more like a marketing headline that keeps everyone in the ecosystem comfortable.

A note has three layers, and they are worth very different amounts.

The rating. Buy, Hold, Sell — sometimes dressed as "Outperform", "Neutral", "Reduce". This is the layer most shaped by incentives, and it is skewed. Across the market, the great majority of ratings are Buy or Hold; genuine Sells are a small minority. Why? A Sell annoys the company's management, and management controls the access — the meetings, the concall answers, the plant visits — that the analyst needs to do her job. A Sell can also sour a banking relationship worth crores. So the honest translation is rough but useful: Buy often means "we like it," Hold often means "we don't, but we can't say so out loud," and an actual Sell is rare enough that it is almost a small act of courage. Read the rating as a weather-vane for incentives, not as advice.

The price target. A single number, usually twelve months out, that looks like a measurement and is actually a guess. It is the output of a model: the analyst assumes a growth rate, a profit margin and a valuation multiple — how many rupees investors pay for each rupee of yearly profit — turns the handle, and a number pops out — ₹1,240. Change any one assumption a little and the target moves a lot. It is also revised constantly, quietly, after the fact, to stay near reality. A confident forecast is a sales aid, not a fact about the future — the more decimal places a target carries, the more certainty it pretends to.

The workings. This is the buried treasure: the model, the assumptions, the segment-by-segment numbers, the industry data, the notes from the last management call. This is what you cannot easily assemble yourself, and it is the reason to open the PDF at all. A good analyst has spent weeks on a company you will spend an afternoon on. Her assumed 15% growth, her margin build-up, her comparison with peers — you can argue with every line of it, and arguing with it is exactly how you learn what actually drives the business.

Walk one note the way you should. illustrative

Imagine a brokerage called Meridian Securities publishes an initiation note — a broker's first published research on a company — on a mid-cap paints company. Top line: BUY, target ₹1,240, current ₹980. Below that, the good stuff. The model assumes revenue grows 15% a year for three years, operating margin — profit from the core business as a share of sales — rises from 16% to 19%, and the stock deserves a multiple of 32 times earnings, meaning investors pay ₹32 for each ₹1 of yearly profit. There is a table splitting revenue into decorative paints and industrial coatings. There is a paragraph noting management said, on the last call, that raw-material costs are easing.

Here is the disciplined read. Skip the ₹1,240 — it is just those three assumptions multiplied out. Instead, poke each assumption. 15% growth for three years: what did the company actually do the last three years, and does 15% assume the good times simply continue? Margin from 16% to 19%: that jump is doing a lot of the work in the target; is there a real reason for it, or is it hope? 32 times earnings: is that what peers trade at, or a generous number chosen to justify a Buy? You have just turned a marketing document into a checklist — and every item on the checklist you can verify in the company's own filing.

Then read the last page, the part in small grey type: the disclosures. If it says Meridian's banking arm managed the company's recent fundraise, or that Meridian holds a position, you have found the thumb on the scale. The note may still be right. But now you know which direction the pressure runs. Take the analyst's numbers back to the company's own filing and you stop being a reader of someone's conclusion and start being a checker of their work.

Even the best note has hard limits, and mistaking its reach for certainty is where readers slip.

It cannot tell you the future price. The target is a forecast, and forecasts of prices twelve months out are, across the whole industry, not much better than coin flips. Precision is not accuracy.

It cannot tell you the analyst's private doubts. What survives into the published note is what is sayable given the firm's relationships. The reservation that would have made a Hold into a Sell often never reaches the page.

And it cannot tell you whether your thesis is right, because it was never written for you. It was written for an institution with a different time horizon, a different risk appetite, and a phone line straight to the analyst that you do not have. You are reading the free, public, downstream version of a private conversation.

The same handful of mistakes catch reader after reader.

1. Reading the rating and stopping. The one word at the top is the most incentive-shaped, least reliable line in the document — and it is the one most people act on. If you only have thirty seconds, spend them on an assumption, not on the verdict.

2. Treating the target as a promise. "Target ₹1,240" is not a forecast the analyst is accountable for; it is quietly revised whenever reality drifts. Anchoring your hopes to a specific number is anchoring to a moving decoration.

3. Counting the Buys. "Fourteen brokers say Buy" feels like fourteen independent votes. It is often one crowded consensus, all leaning the same way for the same structural reasons — and consensus that agrees too neatly should raise suspicion, not comfort.

4. Skipping the disclosures. The small grey type at the back is where the banking relationship and the firm's own holdings are admitted. It is the single most honest page in the note, and almost nobody reads it.

5. Over-correcting into cynicism. The opposite error: deciding all research is garbage and ignoring genuinely useful modelling. That throws away the one part that was worth having.