Your First Annual Report, Guided

Phase 0 · Orientation & Foundations · Week 3 · ≈12 hours of module work

Where this sits. The final module of Phase 0. Behind you are the map of the program and the five questions, how the machinery of markets works, the numeracy bootcamp, and your toolkit and knowledge system. Ahead lies Phase 1, thirteen weeks of accounting, the language of business. This week is the hinge between them. Before you spend seventy-seven weeks learning to read annual reports, you spend one week learning your way around them. You will tour one full Indian annual report and one full US 10-K end to end. For every section you learn what it is FOR and who writes it. You also learn what an analyst eventually extracts from it. Then you write two deliberately naive one-page notes that you will seal, date, and re-grade at week 80. For this week, one rule: navigation, not judgment. You are not yet equipped to judge, and pretending otherwise would teach you bad habits. You are here to learn the building before you learn the tenants.

Learning objectives

You will be able to:

1. Download, unaided and in under ten minutes, the latest annual report of a large Indian listed company (from the company site, BSE/NSE, or screener.in) and the latest 10-K plus DEF 14A proxy of a US listed company (from SEC EDGAR).
2. Name every major section of an Indian annual report and of a US 10-K, in the order it appears, and state in one sentence what each section is for and who writes it.
3. Explain the difference between standalone and consolidated financial statements, why Indian companies publish both, and which one an analyst reads for what.
4. Locate on demand, in either filing: the auditor's opinion and the KAMs/CAMs, the segment note, the related-party note, contingent liabilities, the shareholding pattern and promoter pledge data (India), and the proxy statement and its compensation tables (US).
5. State what an analyst eventually extracts from each section and which later module builds that skill (e.g., "the notes get judged when accounting policy and quality of earnings arrive; the proxy gets judged with capital allocation and the deep dive").
6. Complete the plain-words worksheet on both filings, describing every section in your own words without evaluating it, and file two sealed, dated one-pagers for re-grading at week 80.

Prerequisites & connections

Builds on. The five questions come first, since today you'll see which physical sections of a filing eventually answer each one. Next comes M0.02, after which you know what SEBI, the SEC, RBI, the exchanges, and a listed company are, and you know a 10-K from an 8-K by name. Then there's the numeracy bootcamp, whose percentages and basic arithmetic are the only math needed here. Last comes the toolkit. Your knowledge system is live, with its company-file template and filing-naming convention plus glossary notes. You also have screener.in and EDGAR bookmarked.

Feeds into. Everything. Phase 1 takes the statements and notes you locate today and teaches you to rebuild them by hand: first the income statement, then the balance sheet, M1.04 the cash flow statement, M1.05 the linkage, M1.06–M1.09 the notes topic by topic, and M1.10 returns to these very two filings for the hand-spreading capstone. Phase 2 teaches you to interrogate what you'll merely wave at today, with M2.05 benchmarking off the segment note and M2.06–M2.08 running forensics on revenue and related parties, then contingent liabilities. M3.07 and M8.04 judge the proxy and Board's Report you'll skim today. M8.01 compresses today's six-hour tour into a 45–60 minute professional skim. And M10.04, in week 80, opens the sealed envelope you file this week. Two of the notes mapped today are owned elsewhere and are pointed at here, not taught. M8.06 runs the document lab on real subsequent-events notes, and M1.10 makes classifying them a step of the hand-spreading capstone.

The contract for the week. You will feel out of your depth in the numbers pages. That is by design. Your job is to describe, locate, and map. "The balance sheet is here, it has these headings, it says total assets of about this much" is the register you want, and it is no conclusion. Every time you feel the itch to judge ("is this good?"), write the question in your journal instead. Phase 1 and 2 exist to answer it.

Core teaching content

2. Get the two documents: do this now, before reading further

This tour is guided, but it isn't hypothetical. Download both filings now, save them per the naming convention you set up (e.g., companies/asian-paints/filings/AP-AR-FY25.pdf), and keep them open as you read.

India: Asian Paints Ltd. It is the textbook first read. The product is intuitively graspable and the history is long and clean. It's also a professionally run promoter company with negligible pledging and a well-organized integrated report. TCS is a fine alternative. It's equally clean and longer, though services are harder to visualize on day one. There are three routes to the report. Try all three once so each becomes familiar:

1. Company website, the primary source: asianpaints.com → Investors → Annual Reports → latest integrated annual report PDF (FY2024-25 or newer; expect roughly 400–500 pages).
2. Exchange portal: bseindia.com or nseindia.com → search "Asian Paints" → company page → annual reports. Labels shift with redesigns, but the path is always company page → disclosures/financials.
3. Aggregator shortcut: screener.in → Asian Paints → Documents tab. It is the fastest in practice, but know the primary routes. Aggregators lag and occasionally mislink.

US: Costco Wholesale Corporation. Its 10-K is famously lean at ~75–85 pages, the business model is vivid, and the numbers hide a structural surprise you'll compute in Worked Example 2. Coca-Cola is a fine alternative. It's clean at ~150 pages, though it has more segments and currency noise.

1. sec.gov → EDGAR company search (sec.gov/edgar/search/): ticker COST (CIK 0000909832).
2. Filter filing type to 10-K. Open the latest in the viewer or download it.
3. Filter to DEF 14A and download the latest proxy too. You will tour it in Section 5.

Costco's fiscal year ends the Sunday nearest August 31, on a 52/53-week retail calendar. Costco files the 10-K in October and the proxy around December, and holds its annual meeting in January. Don't be alarmed when one year has 53 weeks. It says so plainly.

How big is all this? The Asian Paints AR runs ~400–500 pages, the Costco 10-K ~80 and the proxy ~60–80. You will NOT read every word. Nobody does, including 30-year veterans. You'll walk every section, read the high-value ones properly, and skim the rest with the worksheet in hand. Budget ~2 hours on the text here, ~4 in the Asian Paints report, ~3 in the Costco 10-K, ~1 in the proxy, ~1 on the one-pagers, and ~1 for practice set and mastery check.

Scale check (illustrative, modeled on ~FY24/FY25 scale; pull the actual filings). Costco's revenue is around $254 bn, or at ~₹83/$ roughly ₹21 lakh crore. Asian Paints' consolidated revenue is around ₹35,500 crore. Costco is roughly 60× larger, yet the anatomy of their filings is almost the same. Learn the skeleton once, and every company becomes navigable.

3. The guided walk: the Indian annual report (Asian Paints)

First, the map. An Indian listed company's annual report is assembled from parts required by two rulebooks, plus voluntary material. The Companies Act, 2013 sets what every company owes its shareholders. SEBI LODR sets what listed companies additionally owe the market. Here is the standard anatomy, in the order you'll usually meet it:

Now walk it, PDF open. For each section: find it, skim it, fill the corresponding row of Worksheet A (Section 6).

(1)–(2) The glossy front and the chairman's letter. Asian Paints, like most large Indian companies, opens with an "integrated report" of value-creation models, "six capitals" diagrams and brand photography. That is Layer 3 on the trust gradient, and nobody audits a flywheel diagram. But read the chairman's/MD's letter in full, once, slowly. It is the one place management speaks in first person.

Ask what they say the company does, and note what went well and badly in their telling. Then ask whether anything negative is acknowledged at all. Years from now you'll grade letters for candour. Today, just notice. For calibration on what a candid chairman's letter sounds like, the Reading list points you to Berkshire Hathaway's, free and the gold standard.

(3) The Board's Report. This is the board's legally mandated account of the year under section 134 of the Companies Act, the AGM speech in writing. Inside you'll find the financial summary and the recommended dividend. Transfers to reserves come next, then the directors' responsibility statement, in which the board formally owns the accounts. Then come particulars of loans/guarantees/investments, energy/technology/forex data, and a chain of annexures:

• AOC-2, covering related-party contracts
• the CSR report on section 135 mandated spending
• managerial remuneration ratios, each director's pay as a multiple of the median employee's, which is India's pay-ratio disclosure
• ESOP details and the secretarial audit report

You can't yet judge any of it. You can note the dividend, spot whether related-party items exist, and record the pay ratios. Pure fact-collection.

(4) The MD&A. LODR requires discussion of industry structure, opportunities and threats, segment performance, outlook, risks and internal controls. Since 2018 it has also required an explanation of any year-on-year change above 25% in key financial ratios. It's management's bridge between the glossy story and the audited numbers. For Asian Paints expect decorative-paint demand, crude-linked raw-material costs, the home-décor push, competitive intensity.

Read it fully. It is plain business English, and the best single section for a beginner to understand what happened this year. Later you'll police it. Does the narrative match the numbers, or explain away what should alarm?

(5) The Corporate Governance Report. LODR mandates it as proof that oversight machinery exists. It covers board composition (executive vs independent), each director's attendance, and the committees with their chairs: Audit, Nomination & Remuneration, Stakeholders' Relationship, Risk, CSR. It also carries remuneration disclosures and the last three AGMs. At the end there's a compliance certificate from a practising company secretary or auditor. Analysts mine it for signals such as boards of promoter friends, directors who skip half the meetings, or an audit committee without a numerate independent chair. Today, count the board, count the independents, check attendance. Facts only.

(6) BRSR. The Business Responsibility and Sustainability Report is SEBI's standardized ESG disclosure, mandatory for the top 1,000 listed companies. Spend five minutes skimming the headings (energy, emissions, workforce, community, conduct) so you know what it covers.

(7)–(8) The fork in the road: standalone vs consolidated. This is the most important structural fact about Indian annual reports, so slow down here.

An Indian company presents standalone statements, covering the parent legal entity alone. Subsidiaries appear in them only as investment line items and dividend income. And, if it has subsidiaries, it presents consolidated statements, which treat the whole group as one economic unit. Every subsidiary's revenues, costs, assets, and debts are folded in line by line, and intercompany transactions are eliminated. You'll get the mechanics with consolidation later. US filers publish only consolidated statements, so there is no standalone set to confuse you.

Why both? Company law thinks in legal entities. The parent's own distributable profits determine the dividend it can pay, and the parent's own balance sheet is what its direct lenders claim against. Economics thinks in groups. If Asian Paints' international subsidiaries or kitchens business make or lose money, the parent's shareholder ultimately owns those results.

So the analyst's default is to read consolidated. Check standalone for dividend capacity and for divergence between parent and group. A healthy parent atop sick subsidiaries is a story, and occasionally a red flag, as the forensics work shows. To see which entities make up the gap, use AOC-1, the one-line-per-subsidiary summary. At illustrative scale (~FY24; pull the actual filing), consolidated revenue ≈ ₹35,500 crore vs standalone ≈ ₹30,700 crore. Roughly ₹4,800 crore is earned in subsidiaries and JVs. That's 13–14% of the group. You'll compute this in Worked Example 1.

A trap to disarm now. Screeners default to one set or the other, and screener.in shows both, with a toggle. From today, every number you write down gets a label, standalone or consolidated. An unlabeled number is a future mistake.

(9) The auditor's reports, plural. Each statement set carries its own independent auditor's report. For Asian Paints the auditor is Deloitte Haskins & Sells LLP (check in your edition). In order, the report contains:

• the opinion, whose clean form reads "…give a true and fair view…"
• the basis for opinion
• Key Audit Matters (KAMs), the issues the auditor judged most significant this year. For a paint company these are typically revenue-linked rebates or discounts, so check yours.
• other information, where the auditor reads the Board's Report and MD&A for inconsistency without auditing them
• the responsibilities paragraphs
• the Report on Other Legal and Regulatory Requirements, which annexes two India-specific gems. One is the CARO 2020 report under the Companies (Auditor's Report) Order, with 21 prescribed checks covering asset records, inventory verification, loans to related parties, statutory dues, defaults to lenders, fraud noticed, and more. Alongside it comes the report on internal financial controls.

Opinions come in four flavours: unmodified or clean, qualified ("except for…"), adverse ("does NOT give a true and fair view"), and disclaimer ("unable to obtain sufficient evidence"). On top of these sit emphasis of matter paragraphs, which flag without qualifying. Today you learn to FIND all this and recognize the vocabulary. Judging it belongs to the accounting and quality-of-earnings work.

One sentence to tattoo somewhere: a clean opinion means the statements fairly present per the framework. It is not a certificate of business quality, honesty, or investability.

(10) The statements themselves. Each set has four faces in Schedule III's standard format. The Balance Sheet shows what the company owns and owes at two year-ends. The Statement of Profit and Loss shows the year's income and expenses, including "other comprehensive income", which you can ignore until the income statement arrives. Then come the Statement of Changes in Equity and the Statement of Cash Flows.

Today, find each one and read the line headings aloud, because the vocabulary should enter through more than your eyes. Note total revenue and total assets. Then note profit for the year and closing cash, and compute nothing else. Each statement gets a full module in Phase 1.

(11) The notes. This is the longest, densest, most valuable part, often 100+ pages across both sets. Note 1 gives corporate information. Note 2 sets out the material accounting policies, which is how this company chooses to measure things. Then there is one note per statement line (share capital, borrowings, PP&E, revenue…). After those come the "back notes" analysts treasure:

• contingent liabilities and commitments, meaning lawsuits and tax disputes plus guarantees, the obligations that might become real
• related-party disclosures, every transaction with promoters, directors, and group entities
• segment information. Asian Paints typically reports paints and home-improvement segments.
• EPS, plus fair-value and financial-risk disclosures
• a Schedule III table of key ratios with explanations for large swings

Today's task is cartography. Find each landmark note and write its number in the worksheet. Phases 1–2 teach what each note can hide.

(11a) The subsequent-events note, and the one distinction that runs it. Somewhere in the back notes, usually short and usually unread, is a note headed "events after the reporting period" (Ind AS 10; in a 10-K, "subsequent events", ASC 855). It exists because a year-end is a date and an annual report is a document, and months separate them. Asian Paints closes on 31 March and the board approves the accounts in May. A US filer closes on 31 December and issues in February. Anything that happens in that gap has to be dealt with somewhere, and this note is where.

The rule has two halves, and an analyst cares about the difference between them. An adjusting event provides evidence about conditions that already existed at the balance-sheet date, so the numbers themselves are changed before they are published. A non-adjusting event reflects conditions that arose after the date. Here the numbers stand, and the event is disclosed in words, with its financial effect estimated where it can be.

Two cases matter to you, one of each kind. Say a customer defaults in April on a receivable outstanding at 31 March. The customer was already failing on 31 March and April merely proved it. So this is adjusting, and the March receivable is written down before publication.

A covenant breach is harder, because the classification turns on which side of the year-end the breach falls. If the company was already in breach at 31 March and the lender agrees to waive it afterwards, conditions at the year-end decide the matter, and long-term debt can be pushed into current liabilities. If the breach itself happens in May on a loan that was compliant in March, the condition arose later and the accounts are not restated.

Non-adjusting events are cleaner. A post-year-end acquisition, a rights issue, a large fundraise or a fire in a plant are all non-adjusting. The balance sheet you are reading is a photograph of a company that had not yet done them, and the note tells you the photograph is out of date.

Read it for two reasons. It is the only part of the report that tells you what has happened since the numbers were struck, and a company that raised debt in May is not the company whose leverage you are about to compute. It also tells you what management chose to call non-adjusting when the classification was arguable, which is a judgment you can grade. The document lab puts real subsequent-events notes in front of you, and the Phase 1 filing capstone makes classifying them a step of its checklist.

(12) Shareholder information & the shareholding pattern. This material appears in or beside the governance report. It includes monthly share-price history vs the Sensex/Nifty and the distribution of holdings. Registrar details are there too. It also includes the table that matters most in India, the shareholding pattern. That table splits ownership between the promoter and promoter group (for Asian Paints ~52–53%, illustrative), foreign portfolio investors, domestic institutions and public. It also shows whether promoter shares are pledged, meaning borrowed against.

Promoter stake and pledge are uniquely Indian vital signs. One measures skin in the game and the other measures borrowing against it. Both are also filed quarterly with the exchanges under LODR Reg. 31, so cross-check the latest quarter on bseindia.com/nseindia.com. Asian Paints' pledge is essentially nil. Later you'll meet companies where it wasn't, and what happened next.

(13) The AGM notice. It is bound in or separate, and it lists the resolutions shareholders vote on. These include adopting the accounts and declaring the dividend. Others appoint directors and auditors, or approve related-party transactions and ESOP schemes. This is the document that turns the annual report into decisions.

The five questions, mapped onto the building. Every analysis answers five questions. This table shows where their raw material physically appears in an Indian AR, and it is the reason navigation is worth a week of your life:

5. The DEF 14A proxy: where the people and the pay live

It is called a "proxy statement" because it solicits your proxy, meaning your authority to vote your shares at the annual meeting without attending. Voting requires informed shareholders, so the SEC packs it with everything about the humans. Analysts read it as attentively as the 10-K. Walk Costco's and locate, in roughly this order:

1. Meeting mechanics and ballot items: directors, auditor ratification, say-on-pay, shareholder proposals.
2. Board of directors: each nominee's bio, age, tenure, independence, committee seats. India carries the equivalent content in the CG report.
3. Compensation Discussion & Analysis (CD&A), the board's essay on how and why executives are paid, covering which metrics, what targets and what peer group. In the capital-allocation work you'll learn why this predicts management behaviour. People do what they're paid to do.
4. Summary Compensation Table, the standardized table of salary, bonus, stock and option awards, incentive pay and total, per named executive, for three years. Find the CEO's total. Just find it.
5. Pay-versus-performance and the CEO pay ratio, meaning CEO comp as a multiple of the median employee's. India's analogue is the Board's Report annexure.
6. Beneficial ownership table: directors' and officers' shares, plus all 5%+ holders. Vanguard and BlackRock appear for nearly every large US company. This is the US cousin of the shareholding pattern, and the structural contrast is worth noting. No "promoter" controls half of Costco, while Asian Paints' promoter group holds ~52–53%. Ownership structure is a first-order fact about power.
7. Related-person transactions and audit fees. Compare audit fees against other services, since a big "other" bucket raises independence questions.

Rosetta stone: India's Board's Report + CG report + remuneration annexures + AGM notice ≈ America's DEF 14A. Same organism, two skeletons, again.

6. Reading words you can't yet judge: the plain-words protocol and the worksheets

You now face ~600 pages you're not yet equipped to evaluate. The failure mode is passive page-turning, with eyes moving and nothing sticking. The fix is a five-rule protocol:

1. One section at a time. Read high-value sections fully and skim the others with purpose.
2. Close the document, then write 1–2 plain sentences beginning "It says…" If you can't, you skimmed too shallowly, so go back. This is the Feynman move, and the whole worksheet is built on it.
3. Log one surprise per section, as in "I had no idea Costco sells gasoline." Surprise is your attention's fingerprint.
4. Log one term you didn't know, and look it up. Then define it in your own words. Your glossary grows by 15+ terms this week: "contingent," "proxy," "consolidated," "accrual"…
5. Describe, never evaluate. Banned words this week: good, bad, cheap, expensive, strong, weak, undervalued. When a judgment forces itself on you, convert it into a journaled question, such as "Is half of operating income from membership fees a lot? How would I even decide?" Questions like that get answered in the quality-of-earnings and business-model work.

Worksheet A: the Indian annual report (Asian Paints). Copy this table into your notes or print it, with one row per section. The "pages" column matters. Editions differ, so you are building the page map of your copy. That map is your navigation exam aid and your companion through the accounting phase.

Worksheet B: the US 10-K + proxy (Costco). Same drill:

7. Same organism, two skeletons: the cross-market map, and a preview of the professional skim

Pin this somewhere. When you need to find X, this is where it lives:

Unique to India: the standalone/consolidated pair, promoter & pledge disclosures, CARO, BRSR, mandated CSR. Unique to the US: Part III incorporation by reference, personal CEO/CFO certifications, the separate proxy, Item 1C cybersecurity.

Preview: how you'll skim in week 64. Today's tour took six-plus hours per country. Later you'll compress it to 45–60 minutes with a fixed order:

Business → Risk Factors (the real ones) → MD&A → the three statements → segment note → cash-flow statement → related-party note → contingent liabilities → auditor's report (India: + CARO) → India: shareholding pattern (promoter % + pledge)

Alongside it runs the Ctrl-F list ("related party," "pledge," "contingent," "going concern," "restated," "exceptional," "guarantee"). After today you already know where every stop on that route is. Speed is just this map plus ~60 weekly reps, and that is what the tour is for.

8. The sealed envelope: filing away your baseline self

The last task is the strangest and, years from now, the one you'll be gladdest you did.

Why. In week 80 you are handed these two companies again and asked to re-grade what you wrote in week 3. Skill growth is invisible day to day. Only a time capsule measures it fairly, and your naive one-pagers are the "before" photo. Writing them also forces retrieval of everything the tour just taught you, which is worth more than three re-readings.

A professional habit hides here too. Recording what you believe before you know better, dated and unedited, is the decision-journal practice (working agreement #6) that will one day keep you honest about investment theses.

The template: one page per company, no more. Write both today, while the filings are fresh:

WHAT I THINK THIS BUSINESS IS · [Company] · [Date] · Sealed until week 80 1. What does this company actually do? (3–4 sentences, my own words, no jargon) 2. Who gives it money, and why do they choose to? (customers: who, and what for) 3. How do I think it makes its profit? (my best guess at the engine) 4. How big is it, and is it growing? (from the words and the two or three numbers I noted) 5. What do I think could seriously hurt it? (2–3 guesses) 6. What surprised me most in the filing? 7. Three questions I cannot answer yet but want to. 8. Confidence in this page, 1–10: ___ Signed · Dated

The rules. One page each. Date and sign. Never edit, not even typos, not even tomorrow. File both in a sealed/ folder (or a literal envelope) inside your knowledge system. Add "OPEN SEALED ONE-PAGERS" to your week-80 calendar and plan, and log one journal line with today's confidence numbers. At week 80 you will mark them like an examiner. You circle factual errors and note naive framings. You list missed drivers, and then you write the "how far I've come" note the master map schedules. Every learner who does this reports the same two feelings on opening day: mild embarrassment, and proof.

Common mistakes & how experts think differently

1. Reading cover-to-cover like a novel. Beginners grind forward from page 1, spending their freshest attention on brand photography. Experts read with a map and a purpose, in nearly the reverse of page order: statements and notes first, gloss last, if ever. You read in page order today only because you were mapping the building. From Phase 1 on, you never will again.

2. Treating the glossy as evidence. It's a managed message, Layer 3 on the trust gradient. Experts take two things from gloss. One is what management wants you to believe. Second, specific claims that can later be tested against audited numbers.

3. Believing a clean audit opinion is a seal of quality. The opinion says the statements fairly present per the framework. It says nothing about business health, valuation, or fraud-proofness. Satyam carried clean opinions into January 2009, and Enron's auditor signed off too. Experts skip the pass or fail and read the auditor's report for texture: KAMs and CAMs, tenure, emphasis paragraphs, CARO exceptions.

4. Quoting an unlabeled Indian number. Pull revenue from a screener without noticing the standalone/consolidated toggle and your analysis is contaminated at the source. Experts label every number with its entity (standalone/consolidated), its period (Indian FY ends March, and Costco's ends near August 31), and its unit (₹ crore vs $m). It costs five characters per number.

5. Confusing the 10-K with the glossy "annual report." The US mailed report is often a letter wrapped around the 10-K. The 10-K is the legal disclosure. In India there is no separate pair, because the annual report IS the statutory filing. Experts ask for the filing and leave the brochure.

6. Risk-factor blindness. Beginners skip Item 1A, or weight all thirty risks equally. Lawyers write it to be exhaustive, so beginners dismiss it as noise. Experts run the G/S filter and diff this year's list against last year's, because new or reworded risks are management telling you something under legal cover.

7. Trying to understand every number now. The itch to compute is admirable and premature, because it is pattern-matching without patterns. Describe, locate, question, don't judge. That is also how experts enter any unfamiliar domain, such as a first insurer or a first REIT. Anatomy comes first and judgment second.

8. Highlighting everything, keeping nothing. A filing with 200 highlights is a filing unread. Experts write one-line summaries in their own words, as the worksheet asks. They file atomic notes. Retrieval beats re-reading, and that is why the close-the-document rule exists.

9. Skipping the proxy because "it's a separate document." Incorporation-by-reference moves the governance and pay content into a different document, and beginners never chase it. Experts read the proxy with the 10-K, always. "Show me the incentive and I will show you the outcome" (Munger; earned properly in the valuation and judgment phases).

10. Anchoring on management's "adjusted" numbers. MD&As and decks feature adjusted EBITDA and core earnings, plus constant-currency growth. Some adjustments are fair and some are flattering. This week, just flag every number labelled "adjusted/underlying/core." The policing toolkit, which reconciles non-GAAP to GAAP, arrives with the accounting-quality work.

Worked examples

All figures below are illustrative, modeled on ~FY24/FY25 scale. Pull the actual filings and recompute with your edition's numbers. What you're learning is the method, and your numbers will differ slightly, as they should.

Worked example 1: The fork in the road: standalone vs consolidated at Asian Paints

Task. Quantify how much of the business is carried outside the parent entity, using navigation and one subtraction.

Step 1: locate. Standalone Statement of Profit and Loss, top line: revenue from operations ≈ ₹30,700 crore. Consolidated Statement of Profit and Loss: ≈ ₹35,500 crore.

Step 2: compute. Gap = 35,500 − 30,700 = ₹4,800 crore. Share outside the parent = 4,800 ÷ 35,500 = 13.5%.

Step 3: identify what's in the gap. Turn to AOC-1. It lists international paint operations (Asia, Middle East, Africa, South Pacific), home-improvement subsidiaries (kitchens, bath fittings) and industrial-coatings JVs. Names only today. No judgment.

Step 4: the reading rule. Consolidated shows everything a shareholder owns, so it is the analyst's default. Standalone is the entity that pays the dividend and services parent debt, so you check it for dividend capacity and divergence.

Answer. About 13–14% of revenue is earned in subsidiaries and JVs. An analyst reading only standalone silently ignores roughly one rupee in seven. Forward pointer: consolidation mechanics, including "non-controlling interest," arrive with consolidation.

Worked example 2: The profit engine in a different line: Costco's membership fees

Task. Using only the Item 8 income statement, uncover Costco's structural secret with two divisions.

Step 1: locate. Two revenue lines (illustrative, ~FY24 scale): net sales ≈ $249.6 bn and membership fees ≈ $4.8 bn, for total revenue ≈ $254.5 bn. Further down, operating income ≈ $9.3 bn.

Step 2: fees as a share of revenue. 4.8 ÷ 254.5 = 1.9%.

Step 3: fees as a share of operating income. Membership fees carry almost no direct cost beyond the card and the database, plus the renewal notice. 4.8 ÷ 9.3 = 52%.

Step 4: plain words. A company with a quarter-trillion dollars of sales earns roughly half its operating profit from a line under 2% of revenue. The merchandise, at razor-thin markups, is structurally the bait that keeps ~76 million member households (illustrative) renewing at ~90%+. It's the renewal cheque that is the profit.

Answer. ≈1.9% of revenue and ≈52% of operating income. The biggest line is not always the profit engine, and business-model work formalizes the idea. Pure navigation plus two divisions produced a real structural insight before you can even read a balance sheet.

Worked example 3: One fact, three rooms: tracing Asian Paints' dividend

Task. Follow one fact, the year's dividend, through the report, to build the cross-referencing habit.

Step 1: the Board's Report. An interim dividend of ≈ ₹5.15 per share was paid and a final of ≈ ₹28.15 is recommended, so the total is ≈ ₹33.30 per share (illustrative, ~FY24).

Step 2: total payout. Share count ≈ 95.9 crore shares, from the share-capital or EPS note. Total = 33.30 × 95.9 crore ≈ ₹3,194 crore.

Step 3: cross-check in the audited statements. The standalone Statement of Changes in Equity shows dividends deducted from retained earnings. Over in the cash-flow statement, the financing section shows dividend cash actually paid. The amounts won't exactly match Step 2, because the final dividend is recommended now, approved at the AGM, then paid next year. Note the timing question and park it until the statements link up.

Step 4: the standalone rule. Dividends come from the parent's own distributable profits. Against standalone profit ≈ ₹5,300 crore (illustrative), payout = 3,194 ÷ 5,300 ≈ 60%.

Answer. ≈₹33.30/share ≈ ₹3,194 crore ≈ 60% of standalone profit. One fact is visible in the Board's Report, the SOCIE, the cash-flow statement, and the AGM notice. Facts appear in multiple rooms, and analysts always cross-check at least two. Forward pointer: payout policy as capital allocation.

Worked example 4: Reading the ownership map: float, and a pledge drill

Task. Extract the ownership structure and free float. Then run a pledge computation on a synthetic company.

Step 1: locate. Open the shareholder-information section and cross-check the latest Reg. 31 quarterly filing on BSE/NSE. Illustrative (~FY24): promoter & promoter group ≈ 52.6%, FPIs ≈ 15%, domestic institutions ≈ 12%, public ≈ 20.4%. Pledged promoter shares: nil.

Step 2: free float. 100 − 52.6 = 47.4% in non-promoter hands. If market cap ≈ ₹2.7 lakh crore (illustrative; check live), float value ≈ 0.474 × 2,70,000 ≈ ₹1.28 lakh crore.

Step 3: pledge drill ("Bharat Fasteners Ltd," synthetic). Promoters hold 55%, and 60% of promoter shares are pledged. Pledged share of the whole company = 0.55 × 0.60 = 33%. In plain words, a third of all shares stand as loan collateral. If the stock falls far enough, lenders can seize and sell, forcing the price lower and triggering more selling. That spiral is why pledge % appears on every Indian analyst's dashboard and in the rapid-analysis Ctrl-F list.

Answer. Asian Paints is ~52.6% promoter-held with ~47.4% float and nil pledge, a stable-promoter structure. In the synthetic drill, 33% of the company pledged is a red flag, treated fully when governance forensics arrives. Today's skill is that you can find and read the table both facts come from.

Worked example 5: Do the four statements tie? A first articulation check (US)

Task. Before you can judge numbers (Phases 1–2), you have to trust that they connect. Take one synthetic company's four statements and verify the three articulation links every real filing must satisfy. It's pure navigation, and you conclude nothing about whether the business is good. The company is Ridgeline Tools Inc. (synthetic), with all figures in $m. Its numbers are internally consistent by construction, because that's what you're testing.

The four faces (synthetic).

• Income statement (Year 1): revenue 1,000 − COGS 600 − SG&A 200 − depreciation 50 = operating income 150; − interest 30 = pre-tax 120; − tax at 25% (30) = net income 90. Dividends declared and paid: 40.
• Cash-flow statement (Year 1): operating +120 (net income 90 + depreciation 50 − receivables 20 − inventory 20 + payables 20); investing −100 (capex); financing −70 (debt repaid 30 + dividends 40); net change −50.
• Balance sheet (two year-ends):

Step 1: the balance sheet balances (both years). Year 0: 900 = 900. Year 1: 940 = 940. Assets = Liabilities + Equity is an identity, so it always holds. If it fails for you, you mis-copied a figure.

Step 2: retained earnings roll forward. Beginning RE 290 + net income 90 − dividends 40 = 340, the same figure the Year 1 balance sheet shows. The income statement's bottom line is the equity statement's key input.

Step 3: cash ties. Opening cash 100 + net change (−50) = 50, which is the "Cash" line on the Year 1 balance sheet. The cash-flow statement is simply that one balance-sheet line, explained.

Answer. All three links hold, so the statements articulate. Look at what you didn't do. You didn't judge whether the profit is good, the debt safe, or the price fair. You only confirmed that the pieces connect, and that's your licence to start reading them. Forward pointer: you rebuild and prove this linkage by hand on real filings at the linkage crown gate of Phase 1.

Worked example 6: Reading the segment note: do the parts add to the whole? (India)

Task. A diversified company shows its divisions in the segment note. Learn to reconcile it. External segment revenues must sum to consolidated revenue, with inter-segment sales disclosed and then eliminated. The company is Bharat Coatings Ltd (synthetic), in ₹ crore, with two reportable segments.

Step 1: locate. Find the segment note. It's one of the landmark "back notes", so write its number in your worksheet. It reports revenue by segment, split into external (third-party) and inter-segment.

Step 2: reconcile. External revenues: 24,000 + 4,000 = 28,000 = consolidated revenue. Segment totals sum to 29,000. Subtract the 1,000 of inter-segment sales, which is Coatings selling to Home Improvement, and you're back at 28,000. Inter-segment revenue nets to zero for the group. It's real internally, but nobody outside bought anything, so leaving it in would double-count.

Step 3: describe the shape (no judgment). Coatings = 24,000 ÷ 28,000 = 85.7% of external revenue, and Home Improvement = 14.3%. One segment dominates. Today that's a fact about structure and nothing more. Whether concentration is strength or fragility is Q2/Q4 work.

Answer. External revenues reconcile to ₹28,000 crore, ₹1,000 crore of inter-segment sales are eliminated, and the group is ~86% one segment. The segment note lets you see a conglomerate's parts, and the first thing to check is simply that the parts add to the whole. Forward pointer: segment-based benchmarking arrives with the ratio work.

Practice set

Work each problem fully before you read its solution (working agreement #3). Problems 1–4 are guided, 5–9 independent, and 10–12 timed, so set a timer where you can see it.

P1 (guided). You want to know what dividend an Indian company's board proposed for the year just ended. Name the section you'd open first, and one other place in the same annual report where the dividend also appears.

Solution. Open the Board's Report, which states the recommended final dividend and any interim already paid. You'll also see the dividend in the Notice of AGM, as the resolution shareholders approve. Cash actually paid shows in the cash-flow statement (financing) and the Statement of Changes in Equity. Any two of those earn full marks. The deeper point is that dividend facts turn up in several rooms: narrative, resolution, and audited statements.

P2 (guided). Which of the following carries a statutory auditor's opinion? (a) the chairman's letter; (b) the MD&A; (c) the consolidated financial statements and notes; (d) the BRSR.

Solution. (c). The auditor opines on the financial statements and notes only. As for the Board's Report and MD&A, the auditor reads them and MD&A for material inconsistency with the accounts, but gives no opinion on them. Nobody checks the chairman's letter and glossy pages in substance. BRSR assurance is an emerging, separate exercise. It isn't the statutory audit opinion. That's the trust gradient, layer by layer.

P3 (guided). Put these 10-K items in the order they appear, and name each: Item 7, Item 1A, Item 8, Item 1.

Solution. Item 1, Business. Item 1A, Risk Factors. Item 7, Management's Discussion & Analysis. Item 8, Financial Statements and Supplementary Data. You've seen these before. They're the first four stops of the professional skim order you previewed: Business → Risks → MD&A → statements.

P4 (guided). In both markets you want every dealing between the company and its insiders (promoters, directors, officers, their entities). Name the specific place(s) you'd look in (a) an Indian AR and (b) US filings.

Solution. (a) India: the related-party disclosures note to the financial statements (both sets), plus AOC-2 annexed to the Board's Report, plus the CG report's disclosures section. (b) US: the related-party note in Item 8 and the "related-person transactions" section of the DEF 14A proxy. Ctrl-F "related party" / "related person" works in all of them. It's a preview of the rapid-analysis speed trick.

P5. An Indian company reports consolidated revenue of ₹52,000 crore and standalone revenue of ₹39,000 crore. What percentage of consolidated revenue arises outside the parent entity, and which schedule lists the entities responsible?

Solution. Gap = 52,000 − 39,000 = ₹13,000 crore. Share = 13,000 ÷ 52,000 = 25%. The entity list is AOC-1, which gives the salient financials of each subsidiary/associate/JV.

Strictly, the subtraction is approximate. A parent may sell to subsidiaries, and intercompany eliminations blur the gap. As a first-pass reading of how much is earned outside the parent, though, this is the standard quick take. Precise mechanics arrive with consolidation.

P6. From a shareholding pattern: promoters 61.2% (no pledge), FPIs 18.3%, DIIs 9.1%, public 11.4%. (a) What is the free float? (b) With market cap ₹1.9 lakh crore, what is the float's value in ₹ crore? (c) Which regulation makes this table appear quarterly on the exchanges?

Solution. (a) Free float = 100 − 61.2 = 38.8%. (b) 0.388 × 1,90,000 crore = ₹73,720 crore. (c) SEBI LODR Regulation 31, under which the quarterly shareholding pattern is filed with the exchanges. Sanity check: FPI + DII + public = 18.3 + 9.1 + 11.4 = 38.8 ✓.

P7. Where exactly do you find (a) the CEO's total annual compensation for a US company, and (b) the ratio of an Indian MD's pay to the median employee's pay?

Solution. (a) The Summary Compensation Table in the DEF 14A proxy. The 10-K's Part III merely incorporates it by reference. (b) The annexure to the Board's Report giving managerial remuneration disclosures, with each director's pay as a multiple of median employee remuneration (Companies Act section 197(12) read with Rule 5). Bonus recall: the US analogue of the multiple is the CEO pay ratio, also in the proxy.

P8. You open a 10-K's Part III and find four near-empty items saying information is "incorporated by reference." Explain, in two sentences, what has happened and what you do next.

Solution. The company has used the SEC's allowance to satisfy Items 10–14 by pointing to its proxy statement, provided the proxy is filed within 120 days of fiscal year-end. So you go to EDGAR, pull the latest DEF 14A, and read the governance, compensation, ownership, related-person, and audit-fee content there.

P9. Give the location of the complete list of subsidiaries in (a) US filings and (b) an Indian annual report, and name one reason an analyst would ever open that list.

Solution. (a) Exhibit 21 to the 10-K, in the Part IV exhibit index. (b) AOC-1, plus the subsidiaries note. Any one reason earns the mark. You might want to see which entities account for the gap between standalone and consolidated, or spot subsidiaries in unusual jurisdictions. You might be hunting related-party counterparties, or trying to understand group structure before reading the consolidation note.

P10 (timed, 15 minutes, your downloaded Costco 10-K). Find and write down: (i) the auditor's name and stated tenure; (ii) the total number of warehouses and the US count; (iii) the two revenue lines and their amounts; (iv) where the share-buyback table lives.

Solution. (i) Item 8, auditor's report: KPMG LLP. The report's final line states tenure ("auditor since 2002" in recent editions; check yours). (ii) Item 1 (and Item 2): ~890 total and ~615 US at FY24's end, though your edition's numbers govern. (iii) Item 8, income statement: net sales ≈ $249.6 bn and membership fees ≈ $4.8 bn (~FY24 scale; write your edition's exact figures). (iv) Item 5, the issuer purchases of equity securities table. With the item map this takes under 15 minutes, and under 5 once the teardown drills have done their work.

P11 (timed, 15 minutes, your Asian Paints AR). Find and write down: (i) how many KAMs the standalone auditor's report lists, and their subject(s); (ii) the names of the reportable segments in the segment note; (iii) the note number for contingent liabilities; (iv) the promoter shareholding percentage.

Solution. Here are the landmark answers at ~FY24 scale. Verify each against your edition. (i) Typically one KAM, around revenue recognition. Specifically, it covers the estimation of rebates and discounts to dealers. (ii) Two reportable segments, paints (by far the larger) and home improvement. (iii) Write the note number from your copy. You'll find it in the "other notes" cluster near related-party and commitments. (iv) ≈ 52–53%, pledge nil. Full marks = correct locations + your edition's exact values recorded in the worksheet.

P12 (timed, 10 minutes, synthetic). "WholesaleCo" (invented) reports: total revenue $260.0 bn, of which membership fees $5.0 bn; operating income $9.6 bn. (a) Membership fees as % of revenue and as % of operating income. (b) If next year fees grow 8% with everything else unchanged, what is the new operating income and its growth rate? (c) One sentence: why does a structure like this make the renewal rate a headline KPI?

Solution. (a) 5.0 ÷ 260.0 = 1.9% of revenue, and 5.0 ÷ 9.6 = 52.1% of operating income. (b) Fee growth = 5.0 × 0.08 = $0.4 bn. Fees carry near-zero direct cost, so operating income ≈ 9.6 + 0.4 = $10.0 bn, and growth = 0.4 ÷ 9.6 = +4.2% from a 1.9%-of-revenue line. (c) Half the profit renews, or doesn't, with the membership base. A small change in renewal behaviour therefore moves profit far more than an equal change in merchandise sales, so the renewal rate is the number that best tracks the profit engine's health. Unit-economics formalization comes with the business-model work.

P13 (guided). Cedar Systems Inc. (synthetic) reports: beginning retained earnings $250m, net income for the year $60m, dividends declared and paid $20m. Its year-end balance sheet shows total assets $770m and total liabilities $330m, with share capital unchanged at $150m all year. (a) Compute ending retained earnings. (b) Compute total shareholders' equity from the balance-sheet identity. (c) Does share capital + your ending retained earnings reconcile with total equity? (d) Name the two statements you'd cross-check the year's dividend figure between.

Solution. (a) Ending RE = 250 + 60 − 20 = $290m. That's the retained-earnings roll-forward: open + profit − dividends. (b) Total equity = assets − liabilities = 770 − 330 = $440m. (c) Yes. Share capital 150 + retained earnings 290 = 440 = total equity, so the statements articulate. Here equity happens to be just these two components, and P18 shows a case where it isn't. (d) The statement of changes in equity, where dividends are deducted from retained earnings, and the cash-flow statement's financing section, which shows dividend cash actually paid. The Board's Report/notes state it too. One fact, cross-checked in at least two rooms: the WE3 habit applied to the audited statements.

P14 (guided). Deccan Cements Ltd (synthetic) reports three segments (₹ crore). Cement: external 18,000, inter-segment 2,000. Ready-Mix Concrete: external 5,000, inter-segment 500. Power (captive): external 1,000, inter-segment 3,000. (a) What is consolidated (external) revenue? (b) State how much inter-segment revenue is eliminated. (c) Give Cement's share of external revenue. (d) Using a single sentence, explain why the captive Power segment's inter-segment figure (3,000) is so much larger than its external one (1,000).

Solution. (a) External revenues: 18,000 + 5,000 + 1,000 = ₹24,000 crore = consolidated revenue. (b) Inter-segment eliminated = 2,000 + 500 + 3,000 = ₹5,500 crore. Check it: segment totals sum to 29,500, less 5,500 = 24,000 ✓. (c) 18,000 ÷ 24,000 = 75%. (d) A captive power plant exists mainly to supply the group's own cement and concrete operations. Most of what it "sells" is internal (inter-segment), and only a little spills to outside customers. That's the pattern the elimination column is there to reveal. Segment benchmarking arrives with the ratio work.

P15. For each item give (i) its trust-gradient layer (1 = audited statements & notes, 2 = regulated narrative, 3 = voluntary gloss) and (ii) whether the statutory auditor issues an opinion on it: (a) the consolidated cash-flow statement; (b) the chairman's letter; (c) the MD&A; (d) the contingent-liabilities note; (e) the "value-creation model" infographic; (f) the corporate governance report.

Solution. (a) Layer 1, yes. It's a face statement. (b) Layer 3, no. Nobody checks it in substance. (c) Layer 2, no. The auditor reads it for consistency, without opining. (d) Layer 1, yes. Notes are part of the audited statements. (e) Layer 3, no. It's pure gloss. (f) Layer 2, no audit opinion. A practising company secretary or auditor gives a separate compliance certificate, and that isn't the statutory audit opinion. The habit to build: before you believe any line, ask who wrote it and who checked it.

P16. Name the audit opinion each situation most likely draws. The choices are unmodified (clean), qualified, adverse, disclaimer, or an emphasis-of-matter paragraph. (a) The statements give a true and fair view in all material respects; (b) one subsidiary's inventory could not be verified and no alternative evidence was available, the amount being material but not pervasive; (c) the auditor could not obtain evidence over most of the group and cannot form any opinion at all; (d) the statements are materially and pervasively misstated; (e) the statements are fair, but the auditor wishes to draw attention to a material uncertainty already disclosed in the notes.

Solution. (a) Unmodified/clean. (b) Qualified ("true and fair except for…"). Material but not pervasive is the textbook trigger. (c) Disclaimer of opinion. The auditor was unable to obtain sufficient evidence, with pervasive possible effects. (d) Adverse. The statements do NOT give a true and fair view. (e) Emphasis of matter. This is a clean opinion plus a paragraph that flags without qualifying, and it changes nothing about the opinion itself. The trap is reading every modification as "adverse". Adverse is reserved for pervasive misstatement, and a disclaimer for pervasive lack of evidence.

P16b (subsequent events). Nilgiri Foods Ltd (synthetic) has a 31 March year-end and its board approves the accounts on 22 May. Classify each of the following as adjusting (the March numbers are changed before publication) or non-adjusting (the numbers stand and the event is disclosed), and say in one clause why. (a) On 14 April a customer owing ₹18 crore at 31 March enters insolvency, its distress having been evident through the March quarter. (b) On 30 April the company breaches a leverage covenant on a term loan it was comfortably compliant with at 31 March. (c) On 9 May the company announces an all-cash acquisition of a competitor for ₹640 crore.

Solution. (a) Adjusting. The condition, an uncollectible receivable, existed at 31 March, and April only supplied the evidence. The receivable and the credit-loss allowance are restated before the accounts are published. (b) Non-adjusting. The breach arose from post-year-end conditions, so nothing on the 31 March balance sheet is restated and the event is disclosed. Now reverse the facts. If the company was in breach at 31 March and the lender waived it in April, the classification flips. The condition existed at the date, and the debt may have to be shown as current. (c) Non-adjusting. The acquisition is a decision taken after the date, so it's disclosed with its financial effect and not booked. The balance sheet you're reading belongs to a company that hadn't yet spent ₹640 crore. For every event in that note, ask whether it revealed something true on the year-end date or created something new after it. That one question decides all three.

P17. Sierra Retail Inc. (synthetic) reports for the year ($m): net sales 5,000; membership fees 100; total revenue 5,100; merchandise cost of sales 4,350; SG&A 500; operating income 250. (a) Express merchandise cost of sales and SG&A as a percentage of total revenue. (b) Compute the operating margin. (c) Express membership fees as a percentage of operating income. (d) One descriptive sentence (no judgment): what does (c) say about where profit concentrates?

Solution. (a) Merchandise cost 4,350 ÷ 5,100 = 85.3%, and SG&A 500 ÷ 5,100 = 9.8%. (b) Operating margin = 250 ÷ 5,100 = 4.9%. (c) 100 ÷ 250 = 40%. (d) A fee line worth under 2% of revenue (100 ÷ 5,100 = 1.9%) supplies 40% of operating income, so profit concentrates in membership and not in merchandise. It's the WE2 pattern, stated descriptively. Common-size only locates where money goes. It is never a verdict, and verdicts are Phases 2–4.

P18 (timed, 12 minutes). Kaveri Foods Ltd (synthetic), consolidated, ₹ crore. From the statements you note: opening cash 600; cash from operating activities +1,400; cash used in investing −1,000; net cash used in financing −300; opening retained earnings 4,200; profit for the year 900; dividends paid 300; year-end total assets 12,000; year-end total equity 7,500; share capital unchanged at 500 all year. Compute: (i) closing cash, and which other statement must show the same figure; (ii) closing retained earnings; (iii) does share capital + closing retained earnings equal total equity, and if not, name one legitimate reason; (iv) year-end total liabilities; (v) to begin answering "what can kill it?" (Q4), name one statement line or note you would read next.

Solution. (i) Closing cash = 600 + 1,400 − 1,000 − 300 = ₹700 crore. It must equal "cash and cash equivalents" on the balance sheet, which is the articulation check of WE5. (ii) Closing RE = 4,200 + 900 − 300 = ₹4,800 crore. (iii) No. 500 + 4,800 = 5,300 ≠ 7,500. The ₹2,200 crore gap is the other components of equity: securities premium, general and other reserves, accumulated other comprehensive income, and, because this is consolidated, non-controlling interest. Equity is more than share capital + retained earnings. (iv) Total liabilities = 12,000 − 7,500 = ₹4,500 crore. (v) Any of these is a reasonable next read. Try the borrowings note for how much debt there is and when it matures, or the contingent-liabilities note for lawsuits and guarantees, plus tax disputes. Customer or segment concentration works too. All of it is Q4 raw material for Phase 2. Scoring: 5/5 in time = fluent navigation. ≤3 = re-do Worked Examples 5–6 before the mastery check.

Applied mini-project: "The Two-Building Tour"

Most of the week goes here (~7–8 of its 12 hours), and this is where the tour turns into something you keep. You'll produce four artifacts that stay in your knowledge system for good, and two of them get re-opened in week 80.

Pull exactly these materials:

1. Asian Paints Ltd, latest integrated annual report PDF: asianpaints.com → Investors → Annual Reports. Cross-check availability on the BSE/NSE company pages and screener.in → Documents. (Swap: TCS, tcs.com → Investors, if you prefer.)
2. Costco Wholesale Corp, latest 10-K and latest DEF 14A: sec.gov/edgar/search/ → ticker COST (CIK 0000909832) → filter by form type. (Swap: Coca-Cola, ticker KO.)
3. The latest quarterly shareholding pattern for Asian Paints from the BSE or NSE company page (the LODR Reg. 31 filing), to cross-check the AR's ownership table.

Deliverables:

• (a) Worksheet A completed, all 13 Indian-AR rows: page map, plain-words summaries, surprises, looked-up terms, landmark note numbers (policies, related-party, contingent liabilities, segments, ratios).
• (b) Worksheet B completed, all 12 US rows, including the Item 1A G/S risk counts and three facts from the proxy (CEO total compensation, total audit fees, board size and number of independents).
• (c) Standalone-vs-consolidated mini-computation for Asian Paints (Worked Example 1 redone with your edition's exact numbers, labeled).
• (d) Glossary, at least 15 new terms defined in your own words, filed as permanent notes.
• (e) Two sealed one-pagers, on the Section 8 template, one per company, dated, signed, filed in sealed/, with the week-80 reminder set.
• (f) One journal entry, with both confidence scores, the section that intimidated you most, and the question you most want answered by Phase 2.

Scoring rubric (self-score after finishing; pass ≥ 15/18, per the program's ≥80% written-work bar):

Nobody grades this but you. That's the condition the rest of your career will run under, too. Score it the way you'd want an employee to score themselves.

Case Lab: run case XL-0.2 in the app's Cases tab.

Reading & resources

• Zerodha Varsity, Module 3, Fundamental Analysis, the "How to Read the Annual Report of a Company" chapter (zerodha.com/varsity). The best free Indian-market walkthrough, and it reinforces today's tour. [Free] [Beginner]
• SEC Investor.gov bulletin, "How to Read a 10-K/10-Q" (investor.gov, search the title). The regulator's own five-page map of the form. [Free] [Beginner]
• SEC, "Beginners' Guide to Financial Statements" (sec.gov). Read after the tour, before Phase 1. A gentle preview of what the four faces mean. [Free] [Beginner]
• SEC EDGAR full-text search (efts.sec.gov via sec.gov/edgar/search/). Practice: search "membership fee" within Costco's filings. [Free] [Beginner tool]
• BSE (bseindia.com) / NSE (nseindia.com) company pages: annual reports, shareholding patterns (Reg. 31), corporate announcements, and screener.in → Documents tab as the fast aggregator. [Free] [Beginner tool]
• Company IR pages: asianpaints.com → Investors; investor.costco.com. Bookmark both. IR pages are always your first primary-source stop. [Free]
• Berkshire Hathaway shareholder letters (berkshirehathaway.com/letters). Read any one recent letter this week purely as calibration, because it's what a candid chairman's letter sounds like. Compare the tone with the letters you read today. [Free] [Beginner]
• Mike Piper, Accounting Made Simple. Read chapters 1–2 this weekend as the on-ramp to M1.01 (the reading spine begins in earnest next week). [Paid] [Beginner]
• Alternates for the tour (if you swap companies, or for a second rep later): TCS integrated annual report (tcs.com → Investors) and Coca-Cola 10-K (EDGAR, ticker KO). [Free]
• For the curious only (skim, don't study): the text of SEBI LODR Schedule V (sebi.gov.in) and Companies Act 2013, section 134 (mca.gov.in / indiacode.nic.in), the two rules that generate most of the Indian AR's statutory sections. [Free] [Advanced]
• Going deeper. Aswath Damodaran's "Understanding Financial Statements" session (pages.stern.nyu.edu/~adamodar → Webcasts / teaching → the intro-accounting class). A deeper, still-free walk through how the three statements fit together, the articulation you met in Worked Example 5, before Phase 1 formalizes it. [Free] [Intermediate]
• Going deeper. Howard Schilit, Jeremy Perler & Yoni Engelhart, Financial Shenanigans. The classic field guide to how the "boring" notes and the cash-flow statement expose manipulation. A preview of the forensic reading you deliberately defer today (Phase 2, M2.06–M2.08). Skim it now for motivation and study it then. [Paid] [Advanced]

The Modern Analyst's Addendum

Everything above teaches this skill from first principles, by hand. That is how you learn it, and the mastery check still tests it that way. This addendum shows how a working analyst amplifies the same skill today. It adds, and it never replaces. (R1/R10)

AI-Augment this skill

{
  "skill": "Navigating an Indian annual report and a US 10-K plus DEF 14A end to end: the three layers of trust, the standalone-versus-consolidated fork, the auditor's report with its KAMs or CAMs and the CARO annexure, the landmark back notes — segments, related parties, contingent liabilities — the shareholding pattern and the pledge line, and the plain-words protocol that turns six hundred pages into a map you can navigate",
  "use": "This is the module where a learner either acquires the habit of opening the document or acquires the habit of asking a machine what it says, and those two people diverge for seventy-seven weeks. So state the boundary first and the techniques second. THE BOUNDARY: a question about the FORM is safe; a question about the FILING is safe only when the filing is in the room; a question about the COMPANY is not safe at all. Three uses sit inside it. (1) NAVIGATION. 'Which Item of a 10-K carries the issuer repurchase table?', 'what is a CARO annexure and what is it annexed to?', 'why do Items 10 to 14 look empty?' — questions about a standardised public form, which is stable, thoroughly documented, and exactly what these tools have read a great deal of. Use them to shorten the hunt, then go to the page. (2) EXTRACTION INTO A TABLE YOU THEN CHECK. With the actual PDF supplied to the tool, ask for the four faces as a table WITH THE PAGE NUMBER OF EVERY FIGURE. The page number is the entire request. A claim without a locator cannot be checked, and a tool that will not give you one has not read the document — it is completing a pattern. (3) TRANSLATION. 'Explain contingent liability / incorporated by reference / non-controlling interest / emphasis of matter in plain words' is a dictionary question, and §6's rule 4 has you defining fifteen-plus terms this week. Draft them this way, then rewrite each in your own words, because the rewriting is the part that is the learning. THE FOURTH THING, which looks like a use and is not: asking what a filing says without the filing present.",
  "tools": ["Chat assistants with the actual PDF attached — Claude, ChatGPT, Gemini — for page-cited extraction and for plain-words translation", "Document and PDF readers that return page locators rather than prose — the locator is what makes the output checkable", "EDGAR full-text search at efts.sec.gov — the deterministic alternative to asking a model whether a filing contains a phrase; it searches the text of filings from 2001 onward and returns the filing", "The filings themselves: asianpaints.com Investors, bseindia.com and nseindia.com corporate filings, screener.in Documents, and sec.gov/edgar for the 10-K and the DEF 14A"],
  "prompt": "I have attached Costco's latest 10-K. Extract the consolidated income statement into a table with one row per line item, one column per fiscal year shown, and a final column giving the PAGE NUMBER in this document where each figure appears. Use the filing's own line-item labels, not standardised ones. Report net sales and membership fees as the two separate lines the filing shows. If a figure I have asked for is not in this document, write 'not in this document' rather than supplying it from anywhere else, and do not compute any ratio, growth rate or comparison — I am doing that.",
  "verify": "A loop, and every step of it is cheap. STEP 1 — OPEN THE CITED PAGE. Every extracted figure, on the page it claims. This is the step that makes the whole workflow legitimate, and skipping it converts a citation into decoration. STEP 2 — RUN THE TIES. Worked example 5's three articulation links plus the balance-sheet identity catch any single transcription slip in a set of statements; the Quantitative lens below shows they catch all thirty of thirty in that table. STEP 3 — LABEL. Entity (standalone or consolidated), period (Indian FY ends in March, Costco's near August 31), unit (₹ crore or $m). §7's labelling rule is the one check the arithmetic cannot perform for you, because a wholly wrong document ties perfectly. STEP 4 — THE NEGATIVE TEST, and do it once, early. Ask the tool for a figure you have already verified with your own eyes and see whether it reproduces it. A tool that gets a number you know wrong has just told you what the rest of its answers are worth, and that is the cheapest calibration available to you. Two situations deserve extra suspicion and both live in this module: SMALL AND MID-CAP INDIAN COMPANIES are thinly represented in any training corpus, so an unsourced figure for one is a plausible shape rather than a recalled fact; and ANYTHING AFTER THE TRAINING CUTOFF cannot be known at all, which includes the most recent annual report — the one §2 just told you to download.",
  "diy": "The gate is unaided. You download the latest Indian annual report and the latest 10-K and DEF 14A in under ten minutes; you name every major section of each in order, say who writes it and who checks it; you explain the standalone-versus-consolidated fork and which one an analyst reads for what; and you locate, on demand and without help, the auditor's opinion and its KAMs or CAMs, the CARO annexure, the segment note, the related-party note, contingent liabilities, the shareholding pattern with its pledge line, and the proxy's Summary Compensation Table. The sealed one-pagers of §8 are written in your own hand from the document, or they are worthless as a before photograph — an AI-drafted page in that envelope makes week 80's comparison a measurement of the tool rather than of you."
}

Modern Data Analysis

By hand first. You downloaded both filings, walked every section in page order, filled Worksheets A and B in your own words, and tied Ridgeline's four statements together. Keep all of it, and keep the page map especially. The "pages in my edition" column of Worksheet A is the most useful artefact the week produces. It's what makes you fast at the teardown, and it exists only if you turned the pages yourself.

Today's workflow, and the two halves are not symmetric. On the US side, the numbers a 10-K prints are also published as tagged XBRL. EDGAR's companyfacts endpoint returns every fact a filer has ever tagged. Each fact carries the accession number of the filing it came from, the period it covers and the US-GAAP tag it was reported under. So a figure arrives with its own citation, and reconciling it becomes a comparison instead of an act of faith. For the same content in bulk, use the quarterly Financial Statement Data Sets. And EDGAR full-text search answers "does this filing say X" deterministically across everything filed since 2001. That's what you actually wanted in the moment you were tempted to ask a chat window.

India is different. There's no free equivalent covering the annual report. The exchanges carry financial results in a tagged format alongside the PDF under LODR. But the Board's Report, the auditor's report with its CARO annexure, the notes, the shareholding pattern and the AGM notice, every section the tour sends you to, exist as the PDF and nothing else. So the Indian workflow is to download the PDF, extract with a tool that returns page numbers, and check each figure on its page.

Write that asymmetry down in your notes. If you don't, the convenience of the US half will set your standard for the Indian half without your noticing. An unnamed asymmetry becomes a permanent quality gap between the two halves of your work, and India is the half where the governance content lives.

Tools & sources (IN + US). Use pandas for the extracted table, requests for the JSON endpoints, and a PDF tool that reports page numbers for everything Indian.

For India, start with the company's own investor-relations page (asianpaints.com → Investors → Annual Reports) as the primary route. Next come the BSE and NSE corporate-filings archives. BSE's annual-report archive is the deepest free one and reaches back into the 1990s, which matters when Phase 8 has you reading a decade oldest-first. screener.in's Documents tab has annual reports and concall transcripts. The quarterly Reg. 31 shareholding-pattern filings on the exchange sites give the promoter and pledge figures the printed report may already have aged out of. And MCA21 at mca.gov.in covers the unlisted group entities that AOC-1 names but doesn't detail.

For the US, use SEC EDGAR at sec.gov/edgar for the 10-K and the DEF 14A, and EDGAR full-text search at efts.sec.gov for phrase hunting across 2001-onward. The companyfacts API and the quarterly Financial Statement Data Sets give you the tagged figures. And the proxy covers everything Items 10 to 14 merely point at.

# The articulation checks as code — the ties from Worked example 5. Recomputed in-session (R3)
B0 = dict(cash=100, recv=120, inv=180, ppe=500, pay=110, debt=300, sc=200, re=290)
B1 = dict(cash=50,  recv=140, inv=200, ppe=550, pay=130, debt=270, sc=200, re=340)
IS = dict(rev=1000, cogs=600, sga=200, dep=50, opinc=150, intr=30, pretax=120, tax=30, ni=90, div=40)
CF = dict(op=120, inv=-100, fin=-70, net=-50)
A, L = ("cash","recv","inv","ppe"), ("pay","debt","sc","re")
ties = {
  "balance sheet balances, both years": all(sum(b[k] for k in A) == sum(b[k] for k in L) for b in (B0, B1)),
  "income statement subtracts down":    IS["rev"]-IS["cogs"]-IS["sga"]-IS["dep"] == IS["opinc"]
                                        and IS["opinc"]-IS["intr"] == IS["pretax"] == IS["tax"]+IS["ni"],
  "cash-flow sections sum":             CF["op"]+CF["inv"]+CF["fin"] == CF["net"],
  "retained earnings roll forward":     B0["re"]+IS["ni"]-IS["div"] == B1["re"],
  "closing cash ties to the balance sheet": B0["cash"]+CF["net"] == B1["cash"]}
print(all(ties.values()))                        # True — the statements articulate
# and the check the ties CANNOT perform: scale every figure to the standalone basis
f = 30700/35500                                  # module's WE1 ratio; 0.8648
print(f"{f:.4f}", f"{35500/30700:.4f}")          # every tie still holds; the margin is 15.6% wrong

Verify. Prove the checker against the worked figures above, then never trust an extraction that hasn't been through it.

• Worked example 1: the standalone-to-consolidated gap is ₹4,800 crore, 13.52% of consolidated revenue, or one rupee in 7.4. An analyst reading only standalone silently discards roughly a seventh of the business.
• Worked example 2: Costco's membership fees are 1.886% of total revenue and 51.61% of operating income.
• Worked example 3: an interim of ₹5.15 plus a final of ₹28.15 is ₹33.30 per share. That's ₹3,193.5 crore across 95.9 crore shares, or 60.25% of a ₹5,300 crore standalone profit.
• Worked example 4: a 52.6% promoter holding leaves a 47.4% float worth ₹1,27,980 crore against a ₹2.7 lakh crore market cap. In the synthetic pledge drill, you land on 33% of the whole company standing as collateral.
• Worked example 5: all eight ties hold. Both years balance at 900 and 940, retained earnings roll 290 + 90 − 40 = 340, and 100 + (−50) = 50 of closing cash.
• Worked example 6: external segment revenues of 24,000 and 4,000 sum to the 28,000 consolidated figure once the 1,000 of inter-segment sales is eliminated, for a mix of 85.71% / 14.29%.

And the scale check from the download step: $254bn at ₹83 is ₹21.08 lakh crore, 59.4 times Asian Paints' ₹35,500 crore. Two companies sit two orders of magnitude apart with almost identical filing anatomy. That's why you learn the skeleton once.

Quantitative lens

The instruction here is navigation, not judgment. It's right. But there's one quantitative skill inside navigation, and the whole addendum turns on it: the arithmetic of checking a number you did not compute. All the arithmetic that follows is exact and was run on the tables above.

The four statements are over-determined, and that redundancy is your instrument. Corrupt any single one of the thirty numbers in Worked example 5's table, one digit and one line, anywhere across the income statement, the two balance sheets and the cash-flow statement. At least one of the eight ties breaks. Thirty out of thirty. One hundred percent.

So a single transcription slip in a set of extracted statements can't hide, whether the slip was yours or a machine's, and catching it costs eight subtractions instead of a second reading. That makes Worked example 5's tie-check far more than politeness toward the method. It's a complete detector for the most common failure mode of any extraction, manual or automated, and it works without your knowing anything about the company.

Think about what it's doing. Reading the table twice checks your attention against itself, and that's the least reliable comparison available. The ties check the table against arithmetic that the preparer had to satisfy and you didn't have to understand.

A break divisible by nine is a transposition, and that's the first thing to look for. Every digit transposition of a number differs from the truth by a multiple of nine. We tested it exhaustively over 59 transpositions of twelve of the figures above, and all fifty-nine held. A single wrong digit lands on a multiple of nine only when a 0 was read as a 9 or a 9 as a 0. Across the same figures that's 28 of 492 substitutions, 5.7%.

So when a balance is out by 18,000 or 4,500 or 900, look for two digits that swapped places before you look for anything else. This is the audit room's old "casting out nines". It survives the move to machines intact, because a model that mis-reads a scanned table transposes the same way a tired human does.

What the ties can't see is the error a whole section here warns you about. Multiply every figure in Worked example 5's table by 0.8648, the standalone-to-consolidated ratio from Worked example 1, and all eight ties still hold, perfectly. Internal consistency proves the numbers came from one document. It can't prove they came from the right one.

Say an extraction silently took the standalone column instead of the consolidated one. It produces a set of statements that balances and rolls forward. It ties too. A margin computed against consolidated revenue would then be understated by a factor of 1.1564, or 13.5% low, and nothing in the arithmetic would object. Your cross-market map disarms that trap with a labelling rule that costs five characters per number. It isn't fussiness. It's the one check the ties are structurally unable to perform on your behalf, so it has to be a habit and not a step.

One more thing is worth measuring on the way past. Costco's membership line supplies 51.61% of operating income from 1.886% of revenue, a profit density of 27.4 times its revenue share. The merchandise line supplies the remaining 48.39% from 98.07% of revenue, a density of 0.49. So per dollar of revenue the membership line is 55.5 times more profit-dense than the merchandise that fills the warehouse. "The biggest line is not always the profit engine" is Worked example 2's sentence, and 55.5 is its magnitude. The ratio itself, profit share divided by revenue share, is a one-line screen you can run on any segment note once benchmarking lets you start comparing.

Now the limits, and the last one matters most. All the figures used here are illustrative, as the worked-examples header says. Recompute them against your edition and expect small differences.

The corruption sweep tests single-cell errors on a table built to tie. A real filing's four faces carry dozens more lines plus reconciling notes, so the same test on a real extraction will catch less than a hundred percent. Correlated errors, such as a whole column read from the wrong year or a misread units row, can pass every tie the way the rescaling above does. And the divisibility rule gives you a candidate explanation for a break. It doesn't find the break itself.

Most important of all, none of this checking says anything about whether the numbers are true. A fraudulent set of financial statements articulates perfectly, because that is what makes it a set of financial statements. Satyam's tied and balanced, and carried a clean audit opinion right up to the confession. Articulation buys you the right to start reading, which is what Worked example 5 says it buys, and it buys nothing more. Phase 1 teaches you to rebuild these statements. Phase 2 teaches you to distrust them. This tour teaches you to find them, and to notice when what you've found isn't what you asked for.

Do it in code. Write the eight ties as assertions and run every extracted table through them before a single figure enters a note. Put an entity-period-unit label in the table itself, where you can't forget it. Check any break for divisibility by nine before hunting anything else. And run the negative test on your tool once, early, on a figure you've already verified. Then go back to the page, because only the filing can tell you what the filing says.

Where this goes next: galaxy cross-links

• Retrieval and RAG, grounding AI in filings and documents (AI0.04). The document-in-the-room requirement turned into an actual method: how a filing gets indexed and retrieved so an answer can carry a locator, and what "grounded" does and does not guarantee once it does.
• Failure Modes, Verification and the primary-source guardrail (AI0.06). The crown of the AI branch, and the full treatment of the two failure modes a first filing tour is most exposed to: the thinly-represented small Indian company, and the filing published after the training cutoff.
• G2 APIs, HTTP, JSON and web scraping (CS2.05). How EDGAR's companyfacts endpoint, the filing index and full-text search actually work, and why a figure that arrives with an accession number attached is a different kind of object from a figure that arrives in a sentence.
• pandas I, data wrangling (DA1.02). The extracted statements as a dataframe carrying its own entity, period and unit columns, so the labelling rule stops depending on your vigilance and starts being enforced by the shape of the data.
• The "So What", synthesis and the answer-first storyline (CN1.03). What comes after navigation. The tour forbids you to conclude anything; that module is where a pile of located facts becomes a defensible answer somebody can act on.

Teach it back & journal

Feynman prompt. Write one page, no jargon, for a smart 15-year-old. Explain what an annual report is. Why does a company have to publish one, why is most of it not checked by the auditor, and how would you find out what the boss gets paid, for an American company and for an Indian one? If any sentence needs a term the 15-year-old wouldn't know, either explain the term in brackets or delete the sentence and try again. File the page in your knowledge system beside the worksheets.

Journal reflection. Copy your two confidence scores (1–10) from the sealed one-pagers. Then write a few frank lines. Which section of either filing intimidated you most, and what specifically about it? Name the single question from your "three questions I can't answer yet" that you most want answered. Finally, predict in one sentence each what week-80 you will find naive about each one-pager. You'll grade these predictions too.

This module's flashcards and mastery quiz are wired into the app: see the node's Quiz and Reviews.

End of the guided read. Phase 0 complete. Next comes the accounting equation and double-entry, where the language lessons begin.