How to Research a Stock

On Wednesday morning, 28 October 2026, between the prep and the lunch rush, Maya reads a headline: L&T's profit is up 33%, and its share has jumped. Since learning how to read an annual report, she can open the filing herself.

A research question has three parts: what is true now, which the filings show; what you assume about later, which is your reading; and what would prove that reading wrong (chapter 8). Mr. Chartist opens a file and writes the question on its label: what does L&T earn, from whom, how well, with what debt and owners, and what does a price of ₹3837.10 assume?

Then confirm the security, because a name is not an address. On 29 October 2026, five shares on NSE had symbols beginning with LT: Larsen & Toubro itself: show the note on Larsen & Toubro · NSE: LT (LT); three listed companies it controls, LTM for IT services (formerly LTIMindtree: show the note on LTM Limited), LTTS for engineering services and LTF, L&T Finance, for loans; and LT Foods (LTFOODS), a different company that sells Daawat rice. The ISIN: show the note on ISIN, a 12-character code given to each security, settles it: L&T's is INE018A01030.

Last, fix the date and the basis. The file records the price's date (the NSE close of Thursday 29 October 2026, ₹3837.10), the filings' dates (27 October 2026 for July to September; 5 May 2026 for the year to March) and the basis: the whole group, consolidated, as how to read an annual report explains. A number without its date cannot be checked again, and the answers change every quarter.

Know who hands the business its money and what they get for it. L&T is paid by whoever needs something big built or made: governments and their companies (sea bridges, metro lines, power lines, water works), oil and gas companies (offshore platforms, refineries), power and renewable-energy developers, and factories. Smaller parts of the group are paid by IT clients (LTM, LTTS), home buyers (L&T Realty) and borrowers (L&T Finance).

A customer signs a contract: that is order inflow. L&T is paid as it builds, over months or years: that is revenue. Work signed but not yet billed is the order book: show the note on Order book. Maya's customers pay at the counter for a burger made in minutes; L&T is paid like a builder, as each floor goes up. The comparison stops at one point: an order can be delayed, changed or cancelled, so it is not money in the bank.

In the year to March 2026, L&T's revenue was ₹285874 crore, 54% of it from outside India. It signed ₹435590 crore of new orders, a record, and its order book at 31 March 2026 was ₹740327 crore, about 2.6 years of that year's revenue (derived).

Now the latest quarter, dated. From July to September 2026 it signed ₹147665 crore of orders, another quarterly record, 63% of them from abroad. By 30 September 2026 its order book was ₹853514 crore: about 2.8 years of the last four quarters' revenue (derived). Of that signed work, 46% is in India and 40% in the Middle East: show the note on of the signed work is abroad.

What it tells you: whose budgets the business depends on (governments and energy companies, in India and the Gulf), and how many years of work are queued. What it does not tell you: how much profit that work will leave. That is the next chapter.

A margin is what is left from each rupee of sales after the costs of earning it. This lesson uses EBITDA: show the note on EBITDA, earnings before interest, tax, depreciation and amortisation: what the operations earn before the cost of borrowing, the wear of machines and tax. An EBITDA margin of 8% means 8 paise kept from each rupee of sales, before those three.

Four years first. L&T's building-and-making businesses (its own name for them is Projects, Products and Manufacturing) sold ₹124046 crore in FY23 and ₹209328 crore in FY26: 69% more in three years (derived). Their EBITDA margin was 8.2%, 7.7%, 7.8% and 7.8%: about 8 paise a rupee, every year.

Then the latest quarter, for the whole group. From July to September 2026, sales rose 16% to ₹78750 crore, but the EBITDA margin was 8.7%, against 10.0% a year before, so operating profit (EBITDA) rose only 0.6% (derived). L&T's own reason is "lower margin across portfolio and change in revenue mix". Management's words are its reading; the margin is the fact.

Why it matters here: on a ₹100 contract at an 8% margin, about ₹8 is left before interest, wear and tax. If dearer steel or a delay on site adds ₹2 of cost, a quarter of that ₹8 is gone (derived, an illustration). Thin margins are not good or bad on their own; they set how little room there is for things to go wrong (chapter 8).

Earnings quality asks three things: does the profit come from the business's ordinary work, will it repeat, and did it turn into cash? It is not an accusation; it only asks what kind of profit it is.

Take the quarter. L&T's profit for its owners rose by ₹1276 crore, from ₹3926 crore to ₹5202 crore. Taking the filing's lines apart (derived): other income, money earned outside the main work, added ₹922 crore, 72% of the rise; lower interest costs ₹290 crore, 23%; the operations themselves (EBITDA) ₹38 crore, 3%; and the rest, small items, a net ₹26 crore.

Other income is: show the note on Other income mostly what L&T earns on its cash and investments. L&T says it rose on "higher group surplus funds & improved yields and one-time gain on divestment" of a business at LTTS. Interest costs fell partly because the Hyderabad Metro, which L&T is selling, is no longer counted in the group.

Nothing here is hidden or wrong: every line is in the filing. But it is a different kind of profit. Interest on a cash pile depends on the pile and on interest rates; a gain on selling a business happens once. A reader who writes "profit +33%" in the note should also write "mostly other income". That answers the opening question: both numbers belong in the note, side by side.

One-offs can push the other way too. For the year to March 2026, L&T reported ₹16084 crore of profit after a one-time provision: show the note on Exceptional item of ₹1155 crore for employee benefits under India's new labour codes, and called ₹17238 crore its "recurring" profit. A company's "recurring" or "adjusted" number is its own measure: note both, and say which one you use.

Last, the cash check. From April to September 2026, L&T's operations, leaving out its lending arm, brought in ₹17990 crore of cash, against ₹9325 crore of profit for its owners. Cash ran ahead of profit, and about half of it, ₹8720 crore, came from working capital: money customers paid ahead and money owed to suppliers. Why profit and cash differ is in how to read an annual report; here we only check the two against each other.

Start in Maya's kitchen. Her customers pay at the counter; her meat and bread supplier sends a bill at the end of the month. Say she buys ₹3 lakh of supplies a month (a made-up figure): for about four weeks, that ₹3 lakh is the supplier's money working in her kitchen. Working capital is the money tied up in day-to-day work: what customers still owe, stock on the shelves and work done but not yet billed, less what the business owes its suppliers and what customers have paid in advance.

L&T's net working capital was 1.7% of revenue at 30 September 2026, against 10.2% a year before: a record low, in L&T's own words. Customers' advances and suppliers' credit fund almost all of its work, which is why cash ran ahead of profit in chapter 4. It can reverse: if advances slow or payments are delayed, cash would lag profit instead.

Debt is one total and two businesses. The group owed ₹136227 crore in borrowings at 30 September 2026. About ₹123310 crore of it, about 90%, belongs to L&T Finance, its lending arm, which borrows to lend: show the note on A lender's borrowing: its loans to customers were about ₹136640 crore. The rest, about ₹12917 crore, sits beside ₹97435 crore of cash and short-term investments across the group. Split the total before judging it, the way an annual report splits the company from the group.

Can the profit carry it? Interest cover: show the note on Interest cover divides operating profit by the interest on borrowing. From July to September 2026, L&T's operating profit (₹6845 crore) was about 14 times its finance costs (₹473 crore) (derived; the lending arm's own interest is counted among its operating costs). L&T's own figure for net debt, borrowing less cash, was a third of its equity. Due dates and terms are in the annual report's notes (how to read an annual report).

At 30 September 2026, L&T's 1375894989 shares were held by mutual funds (19.39%), insurance companies (19.27%), foreign portfolio investors (19.18%), the L&T Employees Trust (14.56%, the largest single holder) and everyone else (27.60%): other institutions (4.23%) and individuals, companies and others (23.37%). Its promoter, the founder or group that controls a company, holds 0%: L&T has none. Who takes part in the market explains promoters.

What that means: no founding family's fortune rides on L&T, and there are no promoter pledges to check (margin, leverage and pledging explains pledges). The board, the large funds and insurers, and the employees' trust do the watching.

Who else shares the profit? L&T controls three listed companies, LTM, LTTS and L&T Finance, that have their own outside shareholders. Of the group's ₹18954 crore profit in FY26, ₹2870 crore, about 15%, belonged to those other owners, called non-controlling interests: show the note on Non-controlling interests, not to L&T's (derived). That is why this lesson says "profit for its owners".

Judge management by actions on record. On 29 April 2026 L&T agreed to sell its whole stake in the Hyderabad Metro to a Telangana government company for ₹1461.47 crore, with completion now expected by 31 December 2026: a decision about where the owners' money is tied up, and one reason its interest costs fell (chapter 4). Read management's own words, such as its presentation's title "A Record Quarter amidst Volatility", as its reading, beside the numbers.

Where to find it: the shareholding pattern is filed with the exchanges every quarter; the governance report sits in the annual report.

One price, one date. L&T closed at ₹3837.10 on NSE on Thursday 29 October 2026. With 1375894989 shares, the whole company was valued at about ₹527945 crore, about ₹5.28 lakh crore (derived).

Price to earnings (P/E) is the price divided by earnings per share (EPS): show the note on Earnings per share (EPS), the profit for owners divided by the number of shares. On the year to March 2026 (EPS ₹116.93) it is 32.8; on the last four quarters (EPS ₹129.86, derived) it is 29.5. Both are right; they use different years, which is why apps show different P/Es: show the note on Forward P/E. Read it as: buyers paid about ₹33 for each ₹1 of the last full year's profit. Its blind side: it takes the profit as given (chapter 4) and ignores debt.

Price to book (P/B) is the price divided by book value per share: what the accounts say belongs to the owners (₹113029 crore at 30 September 2026) divided by the shares, ₹821.49 (derived). L&T's P/B is 4.67. Its blind side: the books miss what no balance sheet records, such as skills, reputation and the order book; P/B says most about businesses whose assets are mostly money, such as lenders.

Dividend yield is the last full year's dividend divided by the price. L&T paid ₹38 a share for FY26, so the yield is 0.99% (derived): owners received about ₹1 in cash for each ₹100 of today's price. Its blind side: a yield rises when a price falls, which can look like a gift when it is a warning. What is the stock market follows that ₹38 into one owner's year.

Compared to what? A ratio needs a yardstick on the same day. NSE's own figures for the Nifty 50 on 29 October 2026 were a P/E of 19.19, a P/B of 2.82 and a dividend yield of 1.42% (NSE's method, not ours). So L&T's price asked more for each rupee of profit and of book than the index's companies did on average. That is a description. It says buyers expected more growth from L&T, or valued its signed work, or both; whether they are right is what the next years decide. Fair comparisons use one basis: the same earnings period, similar businesses, and the company's own past.

A fourth ratio puts debt back in. Enterprise value (EV) is what it would take to own the whole business: the shares' value, plus borrowings, less cash (market cap and stock categories). For L&T it was about ₹566737 crore; divided by the last four quarters' EBITDA (₹28988 crore, derived), that is an EV/EBITDA of about 19.6. It lets two companies with different borrowing compare more fairly. Its blind side here: nine-tenths of L&T's borrowing is its lending arm's (chapter 5), and EV fits lenders badly, so for L&T this ratio is rough.

What the price assumes. Turn the P/E around: if L&T's earnings per share stayed at ₹129.86, ₹3837.10 would be 29.5 years of today's earnings. If they grew 10% a year for five years, the same price would be 18.3 times the fifth year's earnings (derived). The growth a price seems to assume goes in the middle column of the note in chapter 9, as an assumption, never as a forecast.

Start from the company's own list. In its results presentation of 27 October 2026, L&T names the risks to its outlook, among them time and cost overruns on contracts, competition at home and abroad, international operations, government policy, and interest and other costs. Keep the ones this file has already met: thin margins (chapter 3), 40% of signed work in the Middle East (chapter 2), profit growth carried by other income (chapter 4) and working capital that could reverse (chapter 5).

Turn each into a line you can check: a measure, today's reading with its date, the level at which the reading would have to be redone, and where and when it is reported. Maya writes four. The levels are hers, made up for this lesson; they are not L&T's targets and not signals to buy or sell. Signed work falls below about two years of revenue (2.8 years at 30 September 2026, reported every quarter). The building-and-making EBITDA margin falls below 7% for a full year (7.8% in FY26, reported each year). Net working capital climbs back above 10% of revenue (1.7% at 30 September 2026, every quarter). Profit for owners grows mainly from other income for a whole year (72% of July to September 2026's rise, every quarter).

Why write them early? Yes Bank, a private Indian bank: show the note on Yes Bank · NSE: YESBANK, in February 2019. On Wednesday 13 February it said the RBI (India's central bank and bank regulator) had found no gap between its own count of bad loans and the RBI's for the year to March 2018. On 14 February its share closed 30.6% higher, at ₹221.25 (derived). On 15 February the bank disclosed a letter from the RBI calling that release a "deliberate attempt to mislead the public", as two news reports quoted it. On 26 April it reported its first quarterly loss since listing, about ₹1507 crore; on 30 April its share closed 29.2% lower, at ₹168.00 (derived).

A note that had said "if the regulator contradicts management's account of bad loans, my reading fails" would have told its owner on 15 February exactly what to check again. The whole road, 2017 to 2020, is in technical vs fundamental analysis.

What a line is not: it is not a stop price (risk management 101 sets those) and not a prediction. It is the point at which the note's middle column, the assumptions, must be rewritten. Write it while you are calm: a reader who writes it after the news tends to move it (trading psychology).

Research ends in one dated page: the question, what I know (each fact with its source), what I assume, what would change my mind, and when I will check again. It explains a reading; it never gives a verdict.

Here is Mr. Chartist's file on L&T, reduced to one page. Every line in the first column has a source; the second column is his reading, and could be wrong; the third says what would make him read again.

Your turn: copy the five headings (question, what I know, what I assume, what would change my mind, next check) for any company you study. A note that fits on one page, with dates, is one you will actually read again. Build your first watchlist turns notes into a routine.

The note has no decision in it, on purpose. Before any money moves, one more question needs an answer: if the reading is wrong, how much could you lose, and how many shares does that allow? That is the next lesson, risk management 101.