How to Research a Stock

'Will this stock go up?' is not a researchable question. Nobody can answer it, and organising your work around it guarantees you will accept whichever answer feels most confident.

Replace it with three questions that can actually be worked on. First: what does this business do, and is it getting bigger or smaller? Second: does the money the business reports as profit actually arrive as cash? Third: what would have to go wrong for this to be a mistake, and how would I notice?

Notice that none of those three mention price. Price enters at the end, as a separate question about what you are being asked to pay — never at the start, where it would bias everything you read afterwards.

Set a boundary before you begin, too. Research expands to fill whatever time you give it. A first pass on one company should take a couple of focused hours, not a fortnight, and should end with a written page whether the answer is yes, no, or not yet.

Every listed Indian company is legally required to publish a great deal about itself, in fixed formats, on fixed dates. That published material is the primary source. Everything else — news articles, videos, screener summaries, Telegram forwards — is somebody's reading of it, and the further you get from the original, the more of the meaning has been lost or bent.

The filings live in four places. The company files results and announcements with the exchanges, so the NSE and BSE websites carry them under the company's own page. The company also publishes them on its investor-relations section, usually alongside an investor presentation and an earnings call transcript. SEBI's site carries offer documents and regulatory orders. And the annual report — the single densest document — is available from both the company and the exchange.

A forwarded message is at the bottom of that ladder for a reason. It carries no date, no source, no author who can be held to it, and no way to check what was left out. Its purpose is usually to make you act, not to inform you.

One practical habit fixes most of this. Whenever you read a claim about a company, ask where the number came from and go one level closer to the source. Doing this three or four times will show you how often a confident headline has quietly rounded, cherry-picked or misdated its evidence.

Before any number, write one sentence: who pays this company, for what, and how often. If you cannot write it without using the company's own marketing words, you do not understand the business yet, and every ratio you calculate afterwards will be meaningless to you.

Say it the way you would to a relative who knows nothing about markets. 'They make brake parts and sell them to car manufacturers under multi-year contracts.' 'They run diagnostic labs and get paid per test by walk-in patients and by hospitals.' 'They lend to small shopkeepers and earn the difference between what they borrow at and what they lend at.' Each of those tells you immediately what to worry about.

That sentence also tells you the shape of the revenue, which matters more than beginners expect. Money that arrives once, on a large order, behaves nothing like money that arrives every month from a subscription, and neither behaves like money that arrives seasonally. A company whose revenue is lumpy will look wildly inconsistent quarter to quarter without anything being wrong.

Finally, name the segments. Most companies of any size do several things, and the annual report breaks revenue and profit down by segment. Often one small segment is producing most of the profit, and the growth story people repeat is attached to a different one.

A single quarter tells you almost nothing. Quarters are affected by seasons, one-off orders, festival timing, monsoon, and accounting choices. Five years of annual numbers, lined up side by side, tells you the direction the business is genuinely travelling in.

Line up three rows: revenue (total sales), operating profit (profit from the core business before interest and tax), and net profit (what is left after every cost, interest and tax). Margin means profit as a percentage of revenue. Then read two things — is revenue growing, and is the margin holding? Growth with a falling margin means the company is buying its growth by cutting prices or absorbing rising costs. That is not automatically bad, but it is a different story from the one usually told.

Work through an illustrative example. Company A's revenue over five years runs ₹800 crore, ₹960 crore, ₹1,150 crore, ₹1,320 crore, ₹1,520 crore. That is roughly 20%, 20%, 15% and 15% growth — a genuinely growing business, revenue up about 90% over the period.

Now the profit line: ₹64 crore, ₹82 crore, ₹92 crore, ₹99 crore, ₹106 crore. Net margin therefore runs 8.0%, 8.5%, 8.0%, 7.5%, 7.0%. Revenue nearly doubled while margin drifted down by a full percentage point. The follow-up question writes itself — is that raw material costs, price competition, or the mix shifting toward a lower-margin segment? The annual report's management discussion usually addresses it directly.

Debt is not automatically dangerous. Companies borrow to build capacity, and cheap debt used well is how a good business grows faster than its own profits allow. The question is never 'is there debt' — it is 'can this be serviced comfortably if things get worse'.

Two numbers answer most of it. Debt-to-equity compares what the company owes to what the owners have put in, and tells you how much of the balance sheet is borrowed. Interest coverage compares operating profit to the interest bill, and tells you how many times over the company can pay its interest from what it earns. Coverage is the more urgent of the two, because interest has to be paid whatever happens.

Take Company A again, illustratively. Debt of ₹420 crore against shareholders' equity of ₹560 crore gives a debt-to-equity of 0.75 — meaning it has borrowed 75 paise for every rupee the owners have in. Operating profit of ₹165 crore against an interest bill of ₹48 crore gives interest coverage of about 3.4 times. Comfortable, but not spacious.

Then read the trend, which is where the useful information hides. Company A's debt has gone from ₹180 crore to ₹420 crore over the same five years — up about 133%, while revenue rose about 90%. Debt is growing faster than the business. That is a legitimate strategy if capacity is being built ahead of demand, and a warning sign if it is funding working capital that keeps expanding. The annual report tells you which.

Profit is an opinion; cash is a fact. That line gets repeated so often it has lost its force, so here is what it actually means. A sale is recorded as revenue when the goods are delivered and the invoice is raised — not when the customer pays. If the customer takes six months to pay, the profit exists on paper long before the money arrives in the bank.

The cash flow statement is where this becomes visible. The line that matters is cash from operating activities: the cash the core business genuinely generated after paying for its day-to-day needs, called working capital (money tied up in unpaid customer bills and unsold stock). Compare it to net profit for the same year. Over several years, cash from operations should track profit reasonably closely.

Company A, illustratively, reported ₹106 crore of net profit in FY5 but generated only ₹41 crore of cash from operations. Across all five years the pattern holds — about ₹443 crore of cumulative reported profit against roughly ₹180 crore of cumulative operating cash, which is 41%. Something is absorbing the money before it arrives.

The receivables line explains it. Company A's trade receivables — money owed by customers — went from ₹150 crore to ₹420 crore. Expressed in days, that is about 68 days of sales outstanding rising to roughly 101 days. Customers are taking a month longer to pay than they used to, and the growth in reported revenue is partly growth in unpaid invoices.

None of this proves wrongdoing. It might be a deliberate choice to win business by offering longer credit. But it changes the question you ask next: how strong are those customers, and what happens to the reported profit if some of them do not pay?

The shareholding pattern is filed every quarter and takes about three minutes to read. It shows how much of the company the promoters — the founding family or controlling group — own, how much sits with institutions, and how much is with the public.

Two things matter for a beginner. First, is promoter holding steady, rising or falling over recent quarters? A promoter group steadily selling down its own stake is not a verdict, but it is a fact worth an explanation. Second, and more urgent: what portion of the promoters' shares has been pledged as collateral for loans?

Pledging works the way a gold loan does. The promoter borrows and gives shares as security. If the share price falls far enough, the lender can sell those pledged shares to recover the loan — which puts a block of supply into the market at exactly the moment the price is already weak. That is why heavy pledging can turn a mild decline into a sharp one.

Illustratively, if a promoter group holds 54% of Company A and 22% of that holding is pledged, then close to 12% of the whole company sits as collateral against somebody's loan. The number is disclosed. It costs you three minutes to look up, and it is one of the highest-value checks available to a beginner.

A company's customers are its real source of income, and the annual report usually discloses whether that income is spread widely or concentrated in a few hands. This is one of the least glamorous checks and one of the most predictive.

The everyday parallel is a tailoring shop. One shop stitches uniforms for two hundred walk-in customers a month. Another stitches uniforms for one large school under a single annual contract. The second may earn more and look more impressive — until the school changes supplier, and the entire business disappears in one letter.

So find the concentration. Illustratively, if Company A's top three customers are 47% of revenue and the largest single customer alone is 24%, then a quarter of the entire business depends on one relationship, one negotiation, and one contract renewal date. That is not a reason to avoid the company. It is a reason to know when the contract comes up, and to treat that date as a scheduled risk event.

Then extend the same question one step out. Are the customers in a single industry or a single geography? A supplier with two hundred customers who are all in one cyclical sector is far less diversified than the customer count suggests, because all two hundred of them get squeezed at the same time.

By this point you have gathered a lot of supportive detail, and you will have started to like the company. That is precisely when the work is most likely to go wrong, because from here every new fact tends to get read as confirmation.

The correction is a pre-mortem. Imagine it is three years from now and this has been a clear mistake. Then write down the reasons — as though they had already happened, not as vague possibilities. 'The largest customer moved to a cheaper supplier and revenue fell 20%' is a pre-mortem. 'There could be competition' is not.

Aim for five specific failure paths, and give each one two labels: how likely you think it is, and whether you would see it coming in the filings. A risk you would spot two quarters early is manageable. A risk that would appear fully formed in one announcement is not, and it deserves a smaller position or none at all.

The pre-mortem also does something the rest of the checklist cannot. It converts a vague feeling of unease into named, checkable items — and those named items become the things you monitor every quarter instead of re-reading the story that made you buy.

Valuation asks one thing: what are you being asked to pay for the earnings you expect? It does not tell you whether a share will rise. It tells you how much has to go right to justify the price on offer.

The most common ratio is the price-to-earnings multiple — the price divided by the earnings per share. Earnings per share (EPS) is the company's yearly profit divided by its number of shares. If a share trades at ₹400 and the EPS is ₹20, the multiple is 400 ÷ 20 = 20. In plain terms, you are paying twenty rupees for every one rupee of last year's earnings. That is the whole idea; the rest is context.

Now do it honestly, as a range rather than a point. Suppose you believe earnings per share two years out will land somewhere between ₹18 and ₹24 — that is a forecast, and it is uncertain. Suppose the market has historically paid somewhere between 15 and 28 times earnings for businesses of this type. The two ranges combined give an implied value between ₹270 and ₹672.

Look at that spread. The high end is roughly two and a half times the low end, and both are defensible from the same set of facts. That is the real lesson of valuation for a beginner: it is not a calculation that produces an answer, it is a way of making explicit which end of a very wide range you are betting on and why.

Two cautions. Different business types need different measures — a bank is usually read on price-to-book (the share price compared with the company's net worth per share), an early-stage business may have no meaningful earnings at all, and a cyclical company looks cheapest at the top of its cycle when earnings are peaking. And never treat a stated valuation as a fact about today; multiples move constantly, and any number you read was true only on the day it was written.

Research tells you what you think a business is worth. The chart tells you what everyone else is currently willing to pay, and where they have changed their minds before. Read it last, using price behaviour only — no indicators, no oscillators, nothing derived.

Three readings are enough. First, where has the price built a base — a stretch of sideways movement where buyers and sellers were roughly balanced? A base is the market agreeing on a price for a while, and its low is a level participants have already defended once.

Second, where is the supply? Look for levels where earlier advances stalled and reversed, ideally on visibly heavier volume. Those are prices at which enough holders decided to sell. They tend to matter again, because the same holders are still there and remember.

Third, what would invalidate the idea in price terms? Pick a level, in advance, whose breach would mean the market disagrees with your reading strongly enough that you want to step aside and re-examine. Write it down before you buy, or you will negotiate with yourself afterwards.

Illustratively: a share bases between roughly ₹480 and ₹520 for about forty candles on the daily chart. Above it, three earlier rallies stalled near ₹560 on expanding volume — that is the supply shelf. Below, a close under ₹470 on rising volume would break the base and invalidate the reading. Note that everything there is described in candles and levels — never in RSI, never in a moving-average crossover, and never in days or weeks.

The output of research is not a decision. It is a page. If you cannot fill the page, you have not finished researching, and buying at that point is guessing with extra steps.

Six lines cover it. What the business does, in one sentence. Why it could be worth more in three years, in three bullets. What has to stay true for that to happen. What would prove you wrong — split into a business invalidation and a price invalidation. How much of your capital this is, and why that number. And what you will do if each named risk actually shows up.

The invalidation line is the one that gives the page its value. Without it, every fall becomes an opportunity to 'average down' and every rise becomes proof of your own judgement. With it, you have already agreed with yourself what disagreement looks like.

Then set a review date and hold the page against the next set of filings rather than against the price. If the numbers you named are still moving in the direction you expected, the price falling is noise. If those numbers have turned, a rising price does not save the thesis. Judge the thesis by the thesis.

Do this for a company you decide against, too. A rejected page costs you an afternoon and gives you a written record of exactly why you said no — which is the fastest way to notice, a year later, that you keep rejecting things for a reason that turns out not to matter.