How Capital Markets Work: India & US

Phase: 0 · Orientation & Foundations · Week: 2 · Estimated hours: ~14 Where it sits: You have just finished the five questions, the ROIC-above-WACC master idea, and the map of the programme. The financial literacy bootcamp runs alongside this one and completes the week. After it comes your toolkit and knowledge system, where many of the websites you meet here get bookmarked and organised, and then your first guided annual report. Suggested hour budget: core teaching ~6.5 h · worked examples ~1.5 h · practice set ~2 h · mini-project ~2.5 h · mastery check ~1 h · flashcards & journal ~0.5 h.

A share price is not announced by anyone. It is left over.

Every price you will ever analyze is manufactured, and the machinery is what this week takes apart. Before you can ask the five questions intelligently, and Question 5 most of all, what is the price implying?, you need to know where prices come from, who is on the other side of the trade, what the referee enforces, and where the primary data lives. Mechanically enough, in each case, that you can compute what happens inside an order book and inside an index by hand.

Learning objectives

You will be able to:

1. Distinguish primary from secondary markets and, for any share sale (IPO fresh issue, offer for sale, rights issue, QIP, buyback, ordinary exchange trade), state exactly who receives the money and whether the share count changes.
2. Trace an IPO end-to-end in both countries, from DRHP → price band → anchor book → three-day ASBA book-building → allotment → T+3 listing in India, and S-1 → roadshow → bookrunner allocation → pricing → greenshoe → 180-day lock-up in the US, then name the analyst-relevant differences.
3. Read and manipulate an order book: given bids, asks, and an incoming order, compute the fill, the average price, the new best bid/ask, the last traded price, and the impact cost, respecting price-time priority.
4. Construct a free-float, market-cap-weighted index from scratch: compute investable weight factors, weights, the index level via a divisor, and the divisor adjustment when a constituent changes. Then state how Nifty 50, Sensex, S&P 500, and Nasdaq-100 each actually do it.
5. Map the participants: promoters, FIIs/FPIs, DIIs, mutual funds and SIP flows, HFTs, market makers, retail. Then say where to find live data on what each group is doing.
6. Explain clearing and settlement on the T+1 cycle in both countries: what a clearing corporation's novation does, what NSDL/CDSL and DTC actually hold, and how record dates and ex-dates work now that both markets settle T+1.
7. Draw the regulatory map: what SEBI does, what the SEC does, where the RBI touches markets, and which regulator/website you go to for any given filing or data point.
8. Locate every primary data source used in this program's market work: NSE/BSE quote and filings pages, SEBI filings, SEC EDGAR, index methodology documents, AMFI/NSDL flow data.

Prerequisites & connections

Builds on: the five questions, and nothing else. What follows is the machinery behind Question 5 and part of Question 3. You also need basic arithmetic, percentages and weighted averages, which the numeracy bootcamp is sharpening in parallel. No accounting is assumed, and none is needed.

Feeds forward:

• The toolkit: every website named here becomes a configured bookmark, and the data-pulling habits start there.
• Your first annual report: you'll navigate filings whose location you learn here.
• Phase 1 (Accounting): share capital, IPO proceeds, buybacks, and ESOPs all appear in the equity section of the balance sheet. Knowing the market-side mechanics first makes the accounting concrete.
• Phase 2–3: market cap, enterprise value, float, and liquidity are the inputs to every ratio and valuation, and all of them are defined by the machinery here. Reverse-DCF ("what is priced in?") presumes you trust how the price got there.
• Phase 6 & 9: participant behaviour (FII/DII flows, retail F&O losses, index-fund mechanics) is raw material for behavioural finance and portfolio process.
• Phase 7: the RBI's market touchpoints previewed here become a module of their own.

Deliberate non-overlaps, and who owns what. Six neighbours own material that stops short here on purpose, and each is named in the teaching where it becomes relevant. AA1.08 owns index construction, the ETF creation-and-redemption mechanism, tracking error and the implementation cost of running an index portfolio. M0.06 owns the retail product decision. That means which wrapper a saver should actually use once the market mechanics are understood, with its costs and its taxes. DV1.01 owns the futures margin arithmetic. There the initial and maintenance amounts are set per contract by the exchange, and the daily mark-to-market settles the gain or loss in cash. DV1.05 owns exchange-traded derivative mechanics: contract specifications, lot sizes, expiry, settlement and the margin regime. M9.02 section 4.7 owns market efficiency in its three forms, along with the documented anomalies and what survived out-of-sample. E11.04 owns the promoter pledge cascade, which applies the margin-call arithmetic taught here to a controlling shareholder who has borrowed against the company's own stock.

Core teaching content

4.1 The one-screen map of a capital market

Strip away the jargon and a stock market is four subsystems:

1. The primary market, where companies (or their existing owners) sell shares to the public and new money changes hands for the first time.
2. The secondary market, where investors trade those shares among themselves, continuously, on exchanges. The company is not a party to these trades.
3. The plumbing: brokers, clearing corporations, and depositories that make sure the buyer gets shares, the seller gets money, and nobody's default poisons the system.
4. The referees, who force disclosure and police conduct: SEBI in India, the SEC in the US, with the RBI governing the money, bond, and currency side in India.

Two display layers sit on top: order books, which manufacture each stock's price second by second, and indices (Nifty 50, Sensex, S&P 500, Nasdaq-100), which compress thousands of prices into one number.

Why do secondary markets exist at all, if the company gets no money from them? Because liquidity is the price of trust. Nobody would hand a company money for 20 years if they couldn't get out next Tuesday.

The secondary market's exit option is what makes the primary market's fundraising possible. Its price then goes to work for the company. That price is the company's report card and its cost of raising the next rupee or dollar. It's the company's acquisition currency too, and in India especially it sets the collateral value of the promoter's pledged shares. Primary and secondary are one machine.

Here's why it matters to you, and it's worth saying early. You're training to judge whether the secondary market's price is right, which is Question 5. And you can't judge a number when you don't know how it was made.

4.10 Where the primary data lives: the master table

Bookmark all of these when you build the toolkit. "Primary" means the source of record. It's what you cite in a memo, and it's what every secondary app is repackaging, whether that's screener.in, stockanalysis.com or your broker's app.

India:

US:

Here's a habit to build this week. For any company you touch, you should be able to reach its primary filings in under a minute. Use the NSE or BSE announcements page for an Indian name and EDGAR for a US one. The guided annual-report read will make you do this, end to end.

Modern Data Analysis: pull one week of bhavcopy and rebuild a stock's week from it. The bhavcopy is the NSE's end-of-day file, one row per security per trading day, published free in the archive section of nseindia.com. It's the primary source that every Indian screener, chart and app is repackaging. Downloading five of them takes ten minutes, and it removes a layer of trust from your data supply for good. Take one liquid stock you care about and pull the five daily files for one week. Each row carries the day's open, high, low, close, previous close, total traded quantity, total traded value and the number of trades. The delivery file, published alongside it, carries the deliverable quantity. From those columns, rebuild four things by hand before you write any code. The week's open is Monday's open and the week's close is Friday's close. Neither is the average of anything. The week's high is the maximum of the five daily highs, and the week's low is the minimum of the five daily lows. People get this part wrong by taking the highest close. The week's traded value is the sum of the five daily traded values. Approximate it as close price times quantity and you'll land within a fraction of a percent, so do it both ways and see what your shortcut costs. The delivery percentage is deliverable quantity divided by total traded quantity, and here the trap is real. A simple mean of the five daily percentages gives a different number from the week's aggregate quantities, because the days carry different volumes. The aggregate is the one that means something. On a realistic week the two answers differ by about a percentage point. You come away with two habits. First, delivery percentage is the cheapest liquidity-and-conviction signal India publishes, and the US doesn't publish it at all. A day where 39 percent of the traded quantity actually moved between demat accounts is a different day from one where 12 percent did and the rest was intraday churn. Second, a file with one row per security per day is a panel. Once you can load one you can load a year. At that point the index arithmetic of section 4.5 stops being a worked example and becomes something you can run on the real constituent list. Later on, the order book and the index get turned into functions. This exercise makes the same move with the data feed underneath them.

Common mistakes & how experts think differently

1. "Somebody sets the price." Beginners imagine the exchange, the company, or "big players" deciding prices. There's no price-setter, only a book of resting orders and a flow of incoming ones. Professionals lean on what follows from that. The last traded price is history, and only the bid, the ask and the depth behind them can actually be traded on. A stock "at ₹250" that you can only buy in size at ₹257 is a ₹257 stock for you.

2. "I bought the IPO, so I funded the company." Only the fresh-issue portion funds the company. In a pure OFS, as with Hyundai India or LIC, the company received nothing. You bought out an existing owner. And if you bought on listing day instead of in the issue, you were in the secondary market all along, so your money went to whoever sold you the shares. Experts read the fresh/OFS split and the objects of the issue section before anything else in an RHP.

3. "Oversubscribed 60x, must be a great business!" Subscription measures demand at the offered price. That tells you about sentiment and sizing, and nothing about value. Hot books are engineered with small floats, anchor glamour and GMP chatter. Hyundai's book was a sedate ~2.4x and the stock listed slightly down. Neither fact says anything about what the business earns over ten years. So skip "how oversubscribed?" and ask "what does this price imply about the future?" That's Question 5, and Phase 3 gives you the tools.

4. "The Nifty rose 1%, so stocks rose today." An index is a float-weighted average dominated by its giants. Six mega-caps can drag the index up while the median stock falls. Experts read breadth, the count of advances against declines, and check which names moved before concluding anything about "the market."

5. Treating index membership as a quality certificate. Committees and formulas pick the biggest investable names. Quality of business doesn't enter into it. Entry brings forced passive buying and exit brings forced selling, all of it mechanical and blind to valuation. Get into the habit of separating flow effects from fundamentals without thinking about it.

6. Ignoring the invisible costs: spread + impact. Retail thinks "brokerage is my cost." On a ₹10 lakh order in a thin small cap, walking the book can cost 1–3% each way, dwarfing any brokerage. Experts size positions against liquidity, meaning average daily value traded and depth, and use limit orders by default. They know a "cheap" illiquid stock quotes a price at which you can't actually trade meaningful size.

7. Ex-date confusion. People buy on the ex-date or record date and expect the dividend. They panic at the mechanical ex-date price drop, or compare returns across an ex-date without adjusting. Under T+1 the rule fits on one line: entitled = settled on the register by record date = bought at least one business day before it.

8. "SEBI/SEC approved it, so it's safe." Both regulators run disclosure regimes. They force the risks to be written down, and they don't vouch for the business, the price or the promoters. The approval stamp means "the documents are complete." It never means "this is sound." So read the disclosures the regulator forced into existence. The bodies are buried in them.

9. Confusing flows with fundamentals. "FIIs are selling, sell!" Flows move prices short-term and tell you about positioning. A business's value, though, is its future cash flows, and shareholder churn doesn't change them. Use flow data to understand why the price moved and who might be wrong. Never use it in place of analysis.

10. Assuming both order books of a dual-listed Indian stock are equally deep. NSE typically carries ~90%+ of the liquidity, and the BSE quote on a mid-cap may be stale or thin. Experts route where the depth is and quote NSE prices for analysis by default.

Worked examples

Worked Example 1: Airbnb's IPO (US, December 2020): the anatomy of a pop

All figures rounded and illustrative; verify against the 424(b)(4) prospectus on EDGAR and first-day price records.

Start with the facts. Airbnb raised its indicated range from $44–50 to $56–60, then priced at $68, above even the raised range. It sold ≈ 50m shares for gross proceeds of ≈ $3.4bn. Next morning, the first trade came at ≈ $146.

Step 1: the underwriters' take. Mega-deal gross spreads run ~2–3.5%, far below the classic 7%. Assume 2.5%. Fees come to ≈ 0.025 × $3.4bn ≈ $85m, and the company nets ≈ $3.3bn.

Step 2: money left on the table = (first trade − offer) × shares = (146 − 68) × 50m ≈ $3.9bn. In plain terms, Airbnb sold for $3.4bn something the market was willing to pay ~$7.3bn for within eighteen hours. Institutions that got allocations received an instant ~115% markup.

Step 3: three readings. Hold all of them at once.

• Cost view: the company left more on the table than it raised. That's a wealth transfer from Airbnb's owners to the bookrunners' clients.
• Uncertainty view: this was December 2020, mid-pandemic, for a travel company. Nobody knew the clearing price, and underwriters price for certainty of execution. Perfection isn't the goal.
• Testable takeaway: the offer price and the day-one price are both negotiated/sentiment prices. Neither is an appraisal. Check the chart for where the stock traded a year later against $146, and take the lesson to heart: for value, you will always have to do your own work.

Worked Example 2: Hyundai Motor India's IPO (October 2024): reading an OFS like an analyst

Rounded figures; verify against the RHP on SEBI and listing-day data on NSE.

Start with the facts. This was India's largest-ever IPO, at ≈ ₹27,900 crore, with a price band of ₹1,865–1,960 and pricing at the cap of ₹1,960. It was a 100% offer for sale. The Korean parent sold ≈ 14.2 crore shares, 17.5% of the company, and there was no fresh issue at all. Subscriptions came in at QIB ≈ 7x, NII ≈ 0.6x and retail ≈ 0.5x, for a total of ≈ 2.4x. The stock listed ≈ 1.3% below issue price and closed day one ≈ 7% down.

Step 1: follow the money. Hyundai Motor Company in Korea received ≈ ₹27,900 crore minus issue expenses. Hyundai Motor India received ₹0. Its share count stayed at ≈ 81.3 crore shares. This transaction financed nothing about the Indian company's capital, capacity or strategy. It was an ownership transfer with a prospectus.

Step 2: implied valuation. ₹1,960 × 81.3 crore shares gives a market cap of ≈ ₹1.59 lakh crore (≈ $19bn). That's the number you'd test against earnings power in Phase 3. The ₹27,900 crore raise isn't.

Step 3: read the book. Institutions at 7x against retail at 0.5x is a signature. The price was set to clear with professionals, and it left no visible "pop" for the listing-gain crowd, with GMP near zero. So the muted listing wasn't a verdict on the business. It was evidence the seller captured full value. Compare Airbnb, where US bookrunners underpriced into a frenzy, while Hyundai's sellers priced to the top of a rule-bound band. It's the same machine with opposite tuning.

Step 4: what an analyst files away. Post-IPO float is 17.5%, with the parent keeping 82.5%. Promoter concentration is high, and the minimum-public-shareholding clock is ticking. Index inclusion becomes possible once the eligibility rules on float, listing history and, for Nifty, F&O availability are met. Put that in your diary as a future flow event, separate from any view on the business.

Worked Example 3: A ₹5,00,000 trade, end to end through the plumbing

Say you buy 100 shares of TCS at ≈ ₹5,000 on NSE on a Tuesday (the price is illustrative). Follow the trade through the plumbing.

Several costs ride on the trade in India. Brokerage is often zero or flat for delivery. STT, the securities transaction tax, applies to both legs of a delivery trade. Then come exchange charges, GST, stamp duty and the SEBI fee. All together they're of the order of 0.1–0.2% round trip for delivery, and spread and impact dwarf them in illiquid names. The toolkit tabulates them. In the US you'd get commission-free retail brokerage and tiny SEC/TAF fees. Your shares would sit in street name at DTC, on the same T+1 timing since May 2024.

Ex-date overlay. Now TCS announces a dividend with a record date of Friday. You bought Tuesday and settled Wednesday, so you're on Friday's register and entitled. Your friend buys Friday morning "for the dividend". That settles Monday, so your friend isn't on the register and isn't entitled. Worse, the stock opened Friday lower by roughly the dividend anyway. Your dividend lands in your bank within the payment timeline. For interim dividends under the Companies Act that's about a month, and the corporate-action page states each date.

Worked Example 4: Index inclusion as forced buying: the arithmetic

Fully illustrative numbers; the method is what you keep.

Suppose "Stock X" is announced as a Nifty 50 entrant at an expected weight of 1.0%. Assets tracking Nifty 50, meaning index funds, ETFs and EPFO flows benchmarked to it, total ≈ ₹4,00,000 crore. That's a plausible order of magnitude as of mid-2026, but verify it against AMFI and fund factsheets.

• Forced passive demand ≈ 1.0% × ₹4,00,000 crore = ₹4,000 crore of buying. It's concentrated near the reconstitution date, with much of it in that day's closing session, to match the index.
• Stock X's average daily traded value is ≈ ₹300 crore. So the passive demand equals ≈ 13 days of normal volume, squeezed into a few days.
• The machinery predicts price strength into inclusion, a volume spike on effective day, and often a fade afterwards. That flow is a one-time stock adjustment and carries no new information about cash flows. An exiting stock suffers the mirror image.

NSE Indices and S&P DJI publish reconstitution announcements weeks in advance. Those announcements move prices before a single index fund has traded, because front-runners trade the anticipated flow. None of it changes what the businesses earn. Once you see this, you'll stop celebrating inclusion as an achievement or mourning exclusion as a failure. Both are plumbing events with predictable, temporary flow signatures.

Worked Example 5: A rights issue and the theoretical ex-rights price (India)

All figures synthetic; the arithmetic is the transferable skill.

Sunniva Cements Ltd (synthetic) has 20 crore shares trading at ₹450. It announces a rights issue of one new share for every four held (1:4), at a subscription price of ₹300. That's a headline "33% discount", and message boards will call it free money. It isn't. A rights issue is a primary-market fundraise, with new shares and cash going into the company, and the price adjusts to erase the apparent gift.

Step 1: how much is raised, and for whom. New shares = 20 ÷ 4 = 5 crore, and money into the company = 5 crore × ₹300 = ₹1,500 crore. It's a fresh issue, so your next question is what that ₹1,500 crore will earn against the hurdle. You'll recognise the master idea there, now with a cash inflow attached. Share count rises from 20 to 25 crore.

Step 2: the theoretical ex-rights price (TERP). After the issue the whole company is worth its old market value plus the new cash, spread over the larger share count: TERP = (20 cr × ₹450 + 5 cr × ₹300) ÷ 25 cr = (9,000 + 1,500) ÷ 25 = ₹10,500 cr ÷ 25 cr = ₹420. So on the ex-date the quoted price mechanically falls from ₹450 to ₹420. That ₹30 drop has nothing to do with news. It's dilution made visible.

Step 3: the value of one right. The right to buy a ₹420 share for ₹300 is worth ₹120 (TERP − subscription = 420 − 300). Four existing shares carry the right to one new share. So per existing share, the value is (₹450 − ₹300) ÷ (4 + 1) = ₹30.

Step 4: three shareholders, one truth. Take a holder of 4 shares, with cum-rights wealth of 4 × ₹450 = ₹1,800. Follow what each choice does to that wealth.

• Subscribes: pays ₹300 for a 5th share and holds 5 × ₹420 = ₹2,100. Net of the ₹300 outlay that's ₹1,800. Made whole.
• Sells the right: keeps 4 shares now worth 4 × ₹420 = ₹1,680, plus ₹120 for the right, which comes to ₹1,800. Made whole.
• Does nothing: holds 4 × ₹420 = ₹1,680 and lets the right lapse. That's ₹1,680, a ₹120 loss, the full dilution, handed over to everyone who did subscribe.

The point. The "discount" in a rights issue is an accounting illusion, and the ex-rights adjustment removes it. You face only two real decisions. (a) Do you want to commit more capital to this business at this price? (b) If not, sell the right, and never let it lapse. The bigger question for you as an analyst goes back to Step 1. ₹1,500 crore just went into the company, so at what ROIC will it be reinvested? A discounted rights issue funding sub-hurdle expansion is value destruction wearing a coupon.

Worked Example 6: A buyback: the primary market in reverse (US)

Synthetic company; net income held constant to isolate the mechanic. A full model nets off the forgone after-tax interest on the cash spent, which Step 2 flags.

Vantage Micro Inc. (synthetic) earns net income of $1,200m on 400m shares trading at $60. So EPS = 1,200 ÷ 400 = $3.00, P/E = 60 ÷ 3 = 20× and market cap = 400m × $60 = $24,000m. The board spends $2,400m of cash to repurchase stock at $60.

Step 1: shares retired. $2,400m ÷ $60 = 40m shares are bought and cancelled, which leaves a new count of 400 − 40 = 360m.

Step 2: EPS after (net income unchanged). EPS = 1,200 ÷ 360 = $3.33, up from $3.00, an accretion of 3.33 ÷ 3.00 − 1 ≈ 11.1%. No product was sold and no margin improved. Profit stays the same and is simply divided among fewer shares. One caveat, though. The $2,400m of cash was presumably earning something, so net that after-tax interest off net income and the accretion shrinks a little. We hold net income flat here to isolate the pure share-count effect.

Step 3: ownership creep. A patient holder of 1m shares owned 1 ÷ 400 = 0.250% of the company before and 1 ÷ 360 = 0.278% after. That's a larger slice of the same business without buying a single extra share. You can see now why a buyback is the "primary market in reverse". A fresh issue creates shares and dilutes, while a buyback retires shares and concentrates.

Step 4: versus a dividend. Pay the same $2,400m out as a dividend and it comes to 2,400 ÷ 400 = $6.00 per share. That's cash in hand with the share count unchanged, and in many regimes it's taxed immediately. The buyback returns the identical cash by concentrating ownership instead, and the holder's tax waits until sale. Neither is automatically better. It turns on price and tax.

The point. A buyback creates value only when the shares are bought below intrinsic value, because the company is then buying $1 of business for less than $1, on behalf of the holders who stay. Bought above value, it's the ROIC mistake in a new costume, with management overpaying with owners' cash and booking "EPS growth" for it. EPS accretion is arithmetic. Value creation is a price judgment. Question 3, capital allocation, is the habit of telling the two apart.

Practice set

Work each problem fully before you read its solution. Problems 1–5 are guided and mirror the teaching examples. Problems 6–10 are independent, and 11–12 are timed. Set a 12-minute timer for that pair.

P1 (guided). Classify each as primary market, secondary market, or "primary-market process with secondary-market economics," and state who receives the money: (a) you buy 50 Reliance shares on NSE; (b) Swiggy's IPO fresh-issue portion; (c) the OFS portion of the same IPO, sold by early VC investors; (d) an Infosys buyback of its own shares via tender offer; (e) an HDFC Bank rights issue; (f) a US company issues shares gradually through an ATM program.

Solution. (a) Secondary. The selling investor gets the money. (b) Primary. The company receives it, because new shares are created. (c) This one is a primary-market process, since it's part of the public issue, with secondary-market economics. Existing shares change hands and the VCs get the money, so the company gets nothing. (d) The reverse of primary issuance. The company pays out cash and extinguishes the tendered shares. Participating shareholders receive the money, and the share count falls. (e) Primary. The company receives the money, and existing holders got the right to subscribe pro-rata. (f) Primary. The company receives the money, selling new shares directly into the secondary market's order book over time.

P2 (guided). An Indian IPO's price band floor is ₹96. (i) What is the highest cap SEBI's rules allow? (ii) Give the lowest permitted cap too. (iii) The band is set at ₹96–112 and the issue is 40x oversubscribed at the cap. Say where the final price will be, and whether the company could have taken ₹120.

Solution. (i) Cap ≤ 120% of floor = ₹115.20 (in practice ₹115, whole-rupee bands). (ii) Cap ≥ 105% of floor = ₹100.80 (≈ ₹101). (iii) Final price = ₹112, the cap, because demand clears there. It can't take ₹120, because the price can't exceed the published cap. Capturing more would have meant filing with a higher band. The other route is letting the secondary market reprice it after listing, and that's what happens on hot listings.

P3 (guided). A stock quotes: best bid ₹542.10 (1,200 shares), best ask ₹542.60 (900 shares). Compute (i) the spread, (ii) the mid, (iii) the spread in basis points of the mid, and (iv) the immediate cost of buying and instantly selling 100 shares at market (ignore fees).

Solution. (i) 542.60 − 542.10 = ₹0.50. (ii) (542.60 + 542.10)/2 = ₹542.35. (iii) 0.50 / 542.35 = 0.000922 → ≈ 9.2 bps. (iv) Buy 100 @ 542.60 and sell 100 @ 542.10, and you lose 0.50 × 100 = ₹50. That's the full spread, the market maker's pay for standing ready on both sides.

P4 (guided). Bids: 1,200 @ ₹542.10; 1,500 @ ₹541.95; 900 @ ₹541.80. Asks: 900 @ ₹542.60; then higher. A market sell for 2,000 shares arrives. Compute (i) the fill schedule and average price, (ii) impact cost vs the ₹542.35 mid, (iii) the new best bid and LTP.

Solution. (i) 1,200 @ 542.10 = ₹650,520, then 800 of 1,500 @ 541.95 = ₹433,560. Total ₹1,084,080 / 2,000 = ₹542.04. (ii) (542.35 − 542.04) / 542.35 = 0.000572 → ≈ 5.7 bps. (iii) Best bid ₹541.95 (700 remaining); LTP = ₹541.95.

P5 (guided). At 10:01, order X: buy 500 @ ₹315.40. At 10:02, order Z: buy 400 @ ₹315.45. At 10:03, order Y: buy 700 @ ₹315.40. A market sell for 800 arrives at 10:04. Who trades, in what order, and what remains?

Solution. Price priority comes first. Z's ₹315.45 is the best bid, so Z fills its full 400. The remaining 400 goes to the ₹315.40 queue in time order, and X, placed at 10:01, fills 400 of its 500. Y fills nothing. Afterwards X has 100 left at 315.40, Y has 700 at 315.40 behind X's remainder, and Z is gone. LTP ₹315.40.

P6. An IPO raises ₹3,000 crore: ₹1,200 crore fresh issue + ₹1,800 crore OFS, all at ₹600/share. Pre-issue share count: 18 crore. Issue expenses are 2.5%, borne proportionately. Compute (i) new shares created, (ii) post-issue share count and the fresh-issue dilution, (iii) net cash to the company, (iv) net cash to selling shareholders.

Solution. (i) Fresh shares = 1,200 cr / 600 = 2 crore. OFS shares = 3 crore, but those already existed. (ii) Post-issue count = 18 + 2 = 20 crore, and dilution = 2/20 = 10% of the post-issue company, now owned by new fresh-issue investors. (iii) Company: 1,200 × (1 − 0.025) = ₹1,170 crore. (iv) Sellers: 1,800 × 0.975 = ₹1,755 crore. Notice what that means. Of ₹3,000 crore of headlines, only 39% became corporate capital.

P7. An IPO's retail portion is 2.1 crore shares; the lot size is 30 shares. Retail receives 21,00,000 applications, each for exactly one lot. (i) How many lots are available? (ii) Work out the oversubscription. (iii) Find your probability of allotment, and how many shares you get if allotted. (iv) Would applying for 3 lots have helped?

Solution. (i) 2.1 crore / 30 = 7,00,000 lots. (ii) 21,00,000 / 7,00,000 = 3x. (iii) In oversubscribed issues, SEBI's rule is to allot at least one minimum lot to as many applicants as possible, by lottery. That gives 7,00,000 winners from 21,00,000, a probability of 1/3 ≈ 33%, and each winner gets 30 shares (worth 30 × issue price). (iv) No. When even one lot each is impossible, every applicant enters the same lottery for one lot, whatever the application size. Larger applications matter only if shares remain after the one-lot round. So now you know why families apply from multiple demat accounts. Each application is a separate lottery ticket.

P8. Build the "Trident Index" (base = 100) from three stocks. P has 80 cr shares at ₹250 with IWF 0.45. Q has 40 cr shares at ₹1,000 with IWF 0.65. R has 500 cr shares at ₹40 with IWF 0.85. Compute (i) each FF mcap and weight, (ii) the divisor, (iii) the index after a day where P +2%, Q −1%, R +5%.

Solution. (i) FF mcaps: P = 80×0.45×250 = ₹9,000 cr, Q = 40×0.65×1,000 = ₹26,000 cr, R = 500×0.85×40 = ₹17,000 cr. Total ₹52,000 cr → weights 17.31% / 50.00% / 32.69%. (ii) Divisor = 52,000 / 100 = 520. (iii) New FF mcaps: P = 36 cr float shares × 255 = 9,180, Q = 26 × 990 = 25,740, R = 425 × 42 = 17,850. The total is 52,770, so the index = 52,770/520 = 101.48, i.e. +1.48%. Cross-check: 0.1731(2) + 0.50(−1) + 0.3269(5) = 0.346 − 0.500 + 1.635 = +1.48% ✓.

P9. Continuing P8: after that close, R's promoter sells a 5% stake into the market, lifting R's IWF from 0.85 to 0.90. (i) Find the new FF mcap of R and of the index basket. (ii) Find the new divisor, with the index unchanged at 101.48. (iii) Which direction must index funds trade R, and why?

Solution. (i) R float shares: 500 × 0.90 = 450 cr → FF mcap = 450 × 42 = ₹18,900 cr (was 17,850). Basket total = 52,770 + 1,050 = ₹53,820 cr. (ii) Divisor = 53,820 / 101.48 = 530.35. The index doesn't move, because no price changed, and the divisor absorbs the float event. (iii) R's weight rose, to 18,900/53,820 = 35.1% from 33.8%. So index funds must buy R to keep replicating, a pure flow effect triggered by the promoter's sale. Look at the chain. The promoter sells, the float rises, and forced passive buying partly absorbs the sale. Real reconstitutions work the same way.

P10. (i) You buy an NSE stock on Friday; when do the shares hit your demat (no holidays)? (ii) A US stock's dividend record date is Thursday. Give the last purchase date that earns the dividend, and the ex-date. (iii) You bought an Indian stock on its record date and the dividend didn't come. Write the one-line explanation you'd give a friend.

Solution. (i) T+1 lands on Monday, since Saturday and Sunday aren't settlement days. (ii) You must be settled by Thursday, so buy by Wednesday. The ex-date is Thursday, the same as the record date under T+1. (iii) "Ownership transfers at settlement, which is one business day after the trade, so buying on the record date settles the day after the register closed, and the seller kept the dividend."

P11 (timed, part of the 12-minute pair). Name the primary source (site + document/page) for each: (a) the risk factors of an upcoming Indian IPO; (b) Apple's latest annual report; (c) an Indian mid-cap's promoter pledge status; (d) which US hedge funds held Nvidia last quarter; (e) today's FII net buy/sell figure in Indian equities; (f) the exact rules by which the S&P 500 picks members; (g) monthly SIP inflows; (h) an insider purchase by a US company's CFO last week.

Solution. (a) DRHP/RHP on sebi.gov.in (Filings → Public Issues), and also on the exchanges' IPO pages. (b) 10-K on EDGAR (sec.gov). (c) The quarterly shareholding pattern plus SAST/pledge disclosures on the NSE/BSE company pages. (d) 13F filings on EDGAR. Free trackers aggregate them, but EDGAR is the primary source. (e) The NSE FII/DII daily activity page, with SEBI/NSDL for granular FPI data. (f) The S&P DJI index methodology PDF on spglobal.com/spdji. (g) AMFI monthly data. (h) Form 4 on EDGAR.

P12 (timed). A stock shows asks 100 @ ₹1,204.00 and 150 @ ₹1,205.00 (best bid ₹1,203.40). (i) Find the average fill for a market buy of 250. (ii) Find the impact cost against the mid. (iii) Separately, the stock carries an 8.0% index weight and rises 1.5% while all other members are flat. What does the index do, and what is the new index level if it stood at 21,500?

Solution. (i) 100 × 1,204 = 120,400 and 150 × 1,205 = 180,750, so 301,150 / 250 = ₹1,204.60. (ii) Mid = (1,203.40 + 1,204.00)/2 = 1,203.70 → (1,204.60 − 1,203.70)/1,203.70 = 0.000748 → ≈ 7.5 bps. (iii) Index return = weight × stock return = 0.08 × 1.5% = +0.12% → 21,500 × 1.0012 = 21,525.8 (+25.8 points).

P13 (guided). Harappa Steel Ltd (synthetic) has 50 crore shares trading at ₹600 and launches a 1-for-5 rights issue at ₹360. Compute (i) the TERP, (ii) the value of the right attaching to one existing share, and (iii) the fresh capital raised for the company. Then (iv) for a holder of 500 shares, find the wealth after the ex-date if she (a) subscribes fully, (b) does nothing, or (c) sells her rights. Say who bears the dilution.

Solution. New shares = 50 ÷ 5 = 10 crore. Capital raised (iii) = 10 cr × ₹360 = ₹3,600 crore. It's a fresh issue, so the cash goes into the company. (i) TERP = (50 × 600 + 10 × 360) ÷ 60 = (30,000 + 3,600) ÷ 60 = ₹33,600 cr ÷ 60 cr = ₹560, so the price drops mechanically from ₹600 to ₹560. (ii) Right per existing share = (600 − 360) ÷ (5 + 1) = ₹40. Put another way, each new share's right is worth 560 − 360 = ₹200, earned by the five shares behind it. (iv) Cum-rights wealth = 500 × 600 = ₹3,00,000. The entitlement is 500 ÷ 5 = 100 new shares costing 100 × 360 = ₹36,000. (a) Subscribes: 600 × 560 − 36,000 = ₹3,00,000, unchanged. (b) Does nothing: 500 × 560 = ₹2,80,000, a ₹20,000 loss (= 500 × ₹40). (c) Sells the rights: 500 × 560 + 100 × 200 = 2,80,000 + 20,000 = ₹3,00,000, unchanged. The shareholders who subscribed capture the dilution in (b). A lapsing holder simply gifts it to them.

P14 (guided). Ridgeline Software Inc. (synthetic) earns net income of $900m on 300m shares trading at $75 (so EPS $3.00, P/E 25×). It buys back $1,500m of stock at $75. Holding net income constant, compute (i) shares repurchased and the new share count, and (ii) new EPS and the accretion percentage. Then give (iii) the per-share dividend that would have returned the same cash, and (iv) the one-sentence capital-allocation caveat you'd attach.

Solution. (i) Shares bought = 1,500 ÷ 75 = 20m, and the new count = 300 − 20 = 280m. (ii) New EPS = 900 ÷ 280 = $3.21, and accretion = 3.21 ÷ 3.00 − 1 ≈ 7.1%. If the P/E holds at 25×, the price becomes 25 × 3.21 ≈ $80.36. (iii) The same cash as a dividend = 1,500 ÷ 300 = $5.00 per share. (iv) "The ~7% EPS lift creates value only if $75 was below Ridgeline's intrinsic value per share, because a buyback above intrinsic value destroys value while flattering EPS." Any equivalent wording scores, as long as it carries the price test.

P15. A profitable company's IPO raises ₹4,000 crore under the standard route. Use the standard quotas (QIB up to 50%, NII at least 15%, retail at least 35%) and the anchor rule (up to 60% of the QIB quota, allotted one day early). Compute (i) the rupee size of the QIB, NII, and retail portions, and (ii) the anchor allotment and the residual QIB book. Then find (iii) the internal NII split between small-HNI and big-HNI. Finally, (iv) suppose the same issue had used the alternative route (QIB ≥ 75%, NII ≤ 15%, retail ≤ 10%). Give the retail portion, and one line on why that route changes who must show up.

Solution. (i) QIB = 50% × 4,000 = ₹2,000 cr, NII = 15% × 4,000 = ₹600 cr, retail = 35% × 4,000 = ₹1,400 cr, and the sum checks out at ₹4,000 cr. (ii) Anchor = 60% × 2,000 = ₹1,200 cr, leaving a residual (net) QIB book of 2,000 − 1,200 = ₹800 cr. (iii) NII splits one-third to two-thirds. Small-HNI (₹2–10 lakh) gets ₹200 cr and big-HNI (> ₹10 lakh) gets ₹400 cr. (iv) Alternative-route retail = 10% × 4,000 = ₹400 cr, with QIB at ₹3,000 cr. With a 75% QIB floor the deal lives or dies on institutional demand, since retail, capped at 10%, can't carry it. So a loss-making issuer must line up anchors and QIBs before it dares open the book.

P16. (i) An NSE stock in the 10% daily price band closed yesterday at ₹250. Give today's upper and lower limits, and say what it's called if the stock opens at the upper limit and can't trade higher. (ii) The S&P 500 closed at 6,000. Give the index levels that trip the 7%, 13%, and 20% market-wide circuit breakers. Say which of those halt trading for 15 minutes and which end the day. (iii) One line: the difference between an individual price band and a market-wide circuit breaker.

Solution. (i) Upper = 250 × 1.10 = ₹275 and lower = 250 × 0.90 = ₹225. Opening stuck at ₹275 is an upper circuit, a freeze with buyers but no sellers. (ii) 6,000 × 0.93 = 5,580 (−7%, Level 1) and 6,000 × 0.87 = 5,220 (−13%, Level 2) each trigger a 15-minute halt, if hit before 3:25 p.m. 6,000 × 0.80 = 4,800 (−20%, Level 3) ends trading for the day. (iii) A price band caps how far one stock can move in a day and freezes only that stock. A circuit breaker halts the entire market on a big index move. Think of one as a single-name guardrail and the other as a system-wide fuse.

P17. An equity index's price level rose from 20,000 to 21,600 over a year, while its constituents paid dividends averaging a 1.4% yield. (i) Find the price-index return. (ii) Find the approximate total return, which is what the Total Return Index captures. (iii) A fund tracking this index returned 9.0% over the year. Did it out- or under-perform, and against which benchmark is that judgment fair? (iv) State the one-line rule.

Solution. (i) 21,600 ÷ 20,000 − 1 = 8.0%. (ii) TRI ≈ price return + dividend yield = 8.0% + 1.4% ≈ 9.4%. Reinvesting properly gives 1.08 × 1.014 − 1 ≈ 9.5%. (iii) Against the price index the fund looks 9.0 − 8.0 = +1.0% ahead. Against the TRI it's 9.0 − 9.4 = −0.4% behind. The TRI is the fair benchmark, because the fund also received the dividends. Crediting it for "beating" a dividend-blind index double-counts income it already earned. (iv) Always benchmark fund and portfolio returns against the Total Return Index.

P18 (timed stretch, 10 minutes). Cobalt Robotics Inc. (synthetic) prices a US IPO of 20m base shares at $25, with the standard 15% greenshoe. The underwriters sell the full over-allotment to investors on day one. (i) Give the greenshoe size in shares, and the total shares sold to investors. (ii) Hot case: the stock trades to $30. Do the underwriters exercise the greenshoe, and what are the company's total shares issued and gross proceeds? (iii) Weak case: the stock trades to $22. Explain how they cover their position, and why that action supports the price. (iv) In the weak case, find the cost to buy the shares in and the gross gain on the oversold block, before the underwriting spread.

Solution. (i) Greenshoe = 15% × 20m = 3m shares. Total sold to investors = 20m + 3m = 23m, which leaves the underwriters short 3m. (ii) Hot: yes. They exercise the over-allotment option and buy the 3m from the company at the $25 offer price, never at $30, to cover the short. The company issues 20 + 3 = 23m shares, and gross proceeds = 23m × $25 = $575m. (iii) Weak: they let the greenshoe lapse and buy 3m shares in the open market at $22 to cover the short. That buying is a block of real demand hitting the book just as the price is sagging, so it cushions the fall. The company issues only the base 20m. (iv) Buy-in cost = 3m × $22 = $66m. They had received 3m × $25 = $75m for the oversold shares, so the gross gain is $75m − $66m = $9m before the deal's spread. That's their reward for bearing stabilisation risk. Score: 4/4 in time = fluent on primary-market plumbing; ≤ 2 = re-read the US bookrunner process and Worked Example 6 before the mastery check.

P19 (guided: long on margin). You buy 800 shares of a ₹950 stock on 50% initial margin, with maintenance margin at 30% and the broker charging 10% a year on the loan. Over the next year the company pays a ₹22 per share dividend and the stock reaches ₹1,150, where you sell. Compute five things. (i) Your own capital, the loan, and the leverage ratio. (ii) The margin-call price, and the percentage fall from ₹950 that reaches it. (iii) Your ending equity and your return on capital. (iv) The return the same trade would have produced unlevered. (v) The ending equity and return had the stock instead fallen to ₹800 by the same date.

Solution. (i) Position = 800 × 950 = ₹7,60,000. Your capital = 50% × 7,60,000 = ₹3,80,000, the loan = ₹3,80,000, and leverage = 1 ÷ 0.50 = 2.0×. (ii) Call price = 950 × (1 − 0.50) ÷ (1 − 0.30) = 950 × 0.50 ÷ 0.70 = ₹678.57, a fall of 28.57%. It's the same 28.57% as in the worked example, because the call price depends only on the two margin percentages, never on the price or the share count. Verify it: at ₹678.57 the position is worth ₹5,42,857, equity is 5,42,857 − 3,80,000 = ₹1,62,857, and 1,62,857 ÷ 5,42,857 = 0.30 ✓. (iii) Interest = 3,80,000 × 10% = ₹38,000, dividends = 800 × 22 = ₹17,600, and the sale = 800 × 1,150 = ₹9,20,000. Ending equity = 9,20,000 − 3,80,000 − 38,000 + 17,600 = ₹5,19,600, and the return = 5,19,600 ÷ 3,80,000 − 1 = +36.74%. (iv) Unlevered = (1,150 − 950 + 22) ÷ 950 = +23.37%. Leverage of 2.0 delivered 36.74 rather than 46.74, and the ten-point gap is the interest. (v) At ₹800, ending equity = 6,40,000 − 3,80,000 − 38,000 + 17,600 = ₹2,39,600, and the return is −36.95% against an unlevered −13.47%. Notice that ₹800 is above the ₹678.57 call price, so you were never called. A stock doesn't have to reach the call price to hurt you. It only has to fall.

P20 (guided: the short side). You short 500 shares of a $46 US stock on 50% initial margin, with maintenance margin at 30%. Over the holding period the borrow fee is 6% a year on the position's initial market value. The stock pays a $0.90 per share dividend, and you cover at $37. Compute five things. (i) The proceeds, your deposit, and the total credit in the account. (ii) The borrow fee and the dividend you owe, and why you owe the second one. (iii) The profit and the return on your deposit. (iv) The margin-call price and the percentage rise that reaches it. (v) One sentence on why the call arrives after a smaller move than it would on an equivalent long.

Solution. (i) Proceeds = 500 × 46 = $23,000, deposit = 50% × 23,000 = $11,500, and total credit held = 23,000 + 11,500 = $34,500. (ii) Borrow fee = 6% × 23,000 = $1,380, and dividend = 500 × 0.90 = $450. Why do you owe the dividend? The shares you sold were borrowed and then delivered to a buyer, who is now the registered holder and receives the company's payment. The lender you borrowed from is still entitled to be made whole, so the payment comes out of your account. It's a real cost of carrying a short, and it makes high-yield stocks expensive to be short of. (iii) Cover cost = 500 × 37 = $18,500, profit = 23,000 − 18,500 − 1,380 − 450 = $2,670, and return = 2,670 ÷ 11,500 = +23.22%. (iv) Call price = 46 × (1 + 0.50) ÷ (1 + 0.30) = 46 × 1.50 ÷ 1.30 = $53.08, a rise of 15.38%. (v) A long's loss is bounded by the price reaching zero, and its cushion shrinks in proportion to the fall. A short's liability grows without limit. Each dollar of rise costs the same equity and enlarges the position the maintenance ratio is measured against. So the ratio decays from both ends, and the same margins call a short after roughly 15% against a long's 29%.

P21 (guided: reading a bhavcopy week). Five days of bhavcopy rows for one synthetic scrip, ₹ per share and shares:

Compute four things. (i) The week's open, high, low and close. (ii) The week's traded value in ₹ crore, using each day's close as the price proxy. (iii) The week's delivery percentage on aggregate quantities. (iv) The simple mean of the five daily delivery percentages, and which of (iii) and (iv) you would put in a note, and why.

Solution. (i) Open = ₹1,840, which is Monday's open and not the lowest open. Close = ₹1,890, Friday's close. High = max(1,871.50, 1,869, 1,855, 1,884, 1,898) = ₹1,898, and low = min(1,828, 1,833, 1,811, 1,830, 1,866) = ₹1,811. The trap on the high is taking the highest close, ₹1,890. That's eight rupees short of where the stock actually traded on Friday. (ii) Daily values in ₹ crore are 1,856 × 12,00,000 ÷ 1e7 = 222.72, then 165.60, 275.70, 206.14 and 151.20, so the week = ₹1,021.36 crore. (iii) Aggregate = 21,50,000 ÷ 55,00,000 = 39.09%. (iv) Daily percentages are 35.00, 40.00, 30.00, 48.00, 49.00, whose mean is 40.40%. Put 39.09% in the note. The mean gives Friday's 8,00,000-share day the same weight as Wednesday's 15,00,000-share day, so it over-weights the quiet, high-delivery sessions. The gap of 1.31 percentage points is small, and it comes entirely from the weighting. A secondary data source makes that sort of silent choice on your behalf all the time.

P22 (guided: routing and price improvement). A US retail investor's app shows an NBBO of 44.28 bid / 44.33 ask. She sends a market order to buy 400 shares. The broker routes it to a wholesaler, which fills the whole order at $44.3128. (i) The spread, and the midpoint. (ii) The price improvement per share against the ask, in cents, and the total in dollars. (iii) That improvement as a percentage of the spread. (iv) How far the fill sits from the midpoint, per share and in total. (v) Where the broker's revenue on this order comes from, and one sentence on what the Indian equivalent of this order looks like.

Solution. (i) Spread = 44.33 − 44.28 = $0.05, and midpoint = (44.28 + 44.33) ÷ 2 = $44.305. (ii) Improvement = 44.33 − 44.3128 = $0.0172 a share, or 1.72 cents. On 400 shares that's $6.88. (iii) 0.0172 ÷ 0.05 = 34.4% of the spread. (iv) The fill is 44.3128 − 44.305 = $0.0078 a share worse than the midpoint, or $3.12 across the order. The wholesaler keeps that for taking the other side of an order it can price as uninformed. (v) The customer paid no commission, so the broker's revenue is the payment for order flow the wholesaler sends it. It's disclosed in the broker's quarterly Rule 606 report. India has no equivalent. There the order goes to the NSE or BSE central book and fills against whatever is resting there at ₹-level prices. It gets no improvement inside the spread, and nobody is paid for the privilege of seeing it first.

Applied mini-project: "The Market Map": one Indian + one US company

Goal: prove you can pull every market-mechanics fact about a real company from primary sources, and read it like an analyst. Choose one NSE large-cap (TCS by default) and one US mega-cap (Apple by default). Give it ~2.5 hours.

Tasks, India (TCS or your pick):

1. On nseindia.com, pull the live quote with market depth. Record the best bid/ask with quantities, the top-5 depth each side, and the LTP. Compute the spread in ₹ and in bps of mid, and the total ₹ value of top-5 depth on each side.
2. From the exchange's shareholding-pattern page, record the latest quarter's promoter %, FII/FPI %, DII %, public %, and any pledge figure. Then write one line on what this ownership map implies for float and control.
3. From the factsheet on niftyindices.com, confirm Nifty 50 membership and record the stock's current index weight. Note the next scheduled reconstitution month.
4. From the NSE/BSE announcements page, locate the latest annual report and the latest quarterly results filing, and record the filing dates. Don't read them yet. You'll do that in the guided annual-report read.
5. Write the settlement line. If you bought 10 shares next Monday, state the demat-credit day. Then, given a hypothetical record date next Thursday, say whether you'd receive the dividend.

Tasks, US (Apple or your pick):

1. On EDGAR, find the latest 10-K, recording its accession date, and the latest 8-K. Then run one full-text search at efts.sec.gov for the company + "risk factors."
2. From stockanalysis.com or your broker, record the bid/ask and compute the spread in bps, verifying tickers on EDGAR. Note shares outstanding from the 10-K cover page against float from any free source, and write one line on the difference.
3. From the S&P DJI factsheet or a fund factsheet (any S&P 500 ETF will do), record the stock's approximate index weight. Check Nasdaq-100 membership too.
4. From EDGAR, find one recent Form 4 for the company and record the insider's name, role, and trade.

Deliverable: one page per company. Cover where it lists and who owns it, and how liquid it is in spread and depth. Then say which indices carry it at what weight, where its filings live, and how a trade in it settles. Close with three sentences on which of the two order books was deeper, and why that might be.

Scoring rubric (0 = missing/wrong, 1 = partial, 2 = complete & correct). You pass at ≥ 11/14.

File both one-pagers in your knowledge system, which the toolkit formalises. They're the seeds of the company files you'll keep for good.

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

Reading & resources

Core (do these):

• Zerodha Varsity, Module 1, "Introduction to Stock Markets," chapters 2–14. It covers the Indian market machine in plain language: IPOs, exchanges, indices, clearing and settlement, corporate actions. [Free] [Beginner]
• NSE Indices, "Nifty 50 Index Methodology" (niftyindices.com → Methodology) and S&P DJI, "S&P U.S. Indices Methodology" (spglobal.com/spdji). Read the eligibility and maintenance sections and skim the rest. They're the primary documents behind the index arithmetic above. [Free] [Intermediate]
• investor.gov (SEC): "How the Markets Work", a crisp US-side refresher on exchanges, orders, and settlement. [Free] [Beginner]
• SEBI investor education portal (investor.sebi.gov.in), which gives the ASBA, IPO and rights mechanics from the regulator itself. [Free] [Beginner]

Reference (bookmark, dip in):

• Nasdaq-100 methodology (indexes.nasdaqomx.com). Check the current weighting and capping rules. [Free] [Intermediate]
• Jay Ritter's IPO statistics (site.warrington.ufl.edu/ritter), decades of underpricing and long-run-return data, and the evidence behind Worked Example 1. [Free] [Intermediate]
• SEBI's studies on individual F&O trader outcomes (sebi.gov.in → Reports & Statistics). Read the latest one end to end. It'll inoculate you for life. [Free] [Beginner]
• NISM Workbook, "Securities Operations and Risk Management" (Series VII), which takes the Indian plumbing of clearing, settlement and corporate actions to exam-prep depth. It's a free PDF from nism.ac.in. [Free] [Intermediate]
• DTCC learning resources (dtcc.com), on how US clearing and street-name custody work. [Free] [Intermediate]

Books (optional depth):

• Larry Harris, Trading and Exchanges: Market Microstructure for Practitioners. Chapters 1–6, on orders, order books and market structures, are the definitive treatment of the order-book material above. Come back to it whenever microstructure matters. [Paid] [Advanced]
• John C. Bogle, The Little Book of Common Sense Investing, chapters 1–5, on why index funds exist and what they imply. Context for everything about indices. [Paid] [Beginner]
• Michael Lewis, Flash Boys, a gripping and contested tour of HFT-era US market structure. Read the rebuttals too. Entertainment with a syllabus attached. [Paid] [Beginner]

Standing data sources: the full master table above, covering NSE, BSE, SEBI, niftyindices, AMFI, the NSDL FPI monitor, RBI, MCA21, EDGAR, S&P DJI, FINRA, DTCC and stockanalysis.com. You'll bookmark and organise them when you build the toolkit.

Going deeper (optional, once the basics are reflexive):

• SEBI ICDR Regulations, 2018, the primary text (sebi.gov.in → Legal → Regulations). The actual rulebook behind the Indian IPO sequence: price-band limits, category quotas, anchor lock-ins, ASBA. Read the book-building and allocation chapters once. Then the IPO mechanics come from the source itself, and you can look up the current numbers whenever a rule changes. [Free] [Advanced]
• Maureen O'Hara, Market Microstructure Theory, the academic foundation beneath the order book: how order flow, dealer inventory, and private information actually set prices. It's denser than Harris's practitioner text. Read the two together for theory plus practice on where a price comes from. [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 to the hand method and never replaces it. (R1/R10)

AI-Augment this skill

{
  "skill": "Reading a limit order book and computing spread, depth and impact cost; index construction, free-float weighting and the divisor; primary versus secondary markets and the fresh-issue/OFS split; clearing, novation and T+1 settlement; the SEBI/SEC/RBI map and the filing alphabet; and knowing which primary source holds which number",
  "use": "This module is a map of machinery and rulebooks, which makes the honest uses unusually easy to name and the dishonest one unusually dangerous. (1) RESTATING A RULEBOOK YOU HAVE ALREADY DOWNLOADED. The Nifty methodology PDF, the S&P 500 methodology PDF and a DRHP are long, public, and structured; with the document actually in front of the tool, 'turn section 4 into an eligibility checklist and cite the page for each line' produces a scaffold you then tick off against the PDF. The document being in the room is the whole condition. (2) THE FILING ALPHABET AS VOCABULARY. '13D versus 13G', 'DRHP versus RHP versus the 424(b)(4)', 'what is novation' — definitional questions about stable public forms, and this is what these tools are best at. (3) DRAFTING THE ARITHMETIC YOU THEN RUN. Ask for the shape of a TERP calculation or an order-book walk as a worksheet, then run it yourself on §4.4's Meridian book and §4.5's three-stock index, where you already know the answers. (4) DISAMBIGUATING TWO MARKETS. 'What is the Indian analogue of a 13F?' is a question about structure, and the §4.10 master table is where you go to confirm the answer.",
  "tools": ["Chat assistants — Claude, ChatGPT — for rulebook restatement with the PDF supplied, for the filing alphabet, and for India/US structural analogies", "Document and PDF tools that return page or section locators — for the methodology documents and the DRHP/RHP", "The primary sources in §4.10 themselves: nseindia.com, bseindia.com, sebi.gov.in, niftyindices.com, sec.gov/edgar, spglobal.com/spdji"],
  "prompt": "I have attached the current Nifty 50 index methodology document. Using only this document, list every eligibility criterion a company must satisfy to be considered for inclusion, and for each one give the section or page number where it appears. Do not add criteria from memory, do not tell me the current constituents, and if a criterion I would expect is not in this document, say that it is absent rather than supplying it.",
  "verify": "One number and one class of fact, and the second is the trap this module is full of. THE NUMBER: never take a level, a weight, a constituent list, a lot size, a fee rate or a subscription figure from a chat window — §4.10 names the source of record for every one of them, and 'the source of record' is what the word primary means. THE CLASS OF FACT: almost everything mechanically important in this module CHANGED RECENTLY. India moved to T+1 in January 2023 and the US in May 2024; India's T+0 window is a phased beta; SEBI retightened the F&O rules across 2024–25; index committees reconstitute twice a year and the S&P size floor is raised periodically. A model's training has a cutoff, and a superseded settlement cycle or margin rule is exactly the kind of fact it will state with complete confidence because the old answer was true for years and is well represented in what it read. This module writes 'verify current' beside these facts for a reason; a chat assistant is the one instrument that cannot perform that verification, because it has no way to know which side of its own cutoff a rule changed on. Go to the exchange, the regulator, or the methodology PDF — all free, all in §4.10.",
  "diy": "The gate is unaided. You walk a limit order book with a pencil and produce the fill schedule, the average price and the impact cost in basis points; you apply price-time priority to a contested queue; you build a free-float index from shares, prices and IWFs, compute the divisor, and adjust it for a constituent change; you compute a TERP and the value of one right; you compute buyback accretion; and you state which regulator and which website holds a given number. No tool is in that room, and none of the above needs one."
}

Modern Data Analysis

By hand first. You walked the Meridian Foods book with a pencil through five scenarios, built a three-stock index from shares, prices and investable weight factors, set its divisor, and adjusted that divisor for a constituent swap. Don't lose that. Anyone can call a function that returns an impact cost. Only someone who has eaten through a book level by level knows that the number is the price of impatience, and understands why it grows.

Today's workflow. Both objects are ten-line functions, and writing them changes what you can ask. The book stops being a table and becomes a walk that takes size as an argument. Impact cost turns into a curve, where the exchange publishes a single number, and you can read your own order size off it before you send it. The index stops being a level and becomes a weight vector plus a divisor. Now you can decompose a day's move into per-stock contributions and measure how concentrated the basket actually is. You can even simulate a reconstitution before the effective date. Both are the same move. You replace a worked instance with a function of the thing that varies.

Tools & sources (IN + US). Use numpy and pandas for the walk and the weights. For India, start with NSE's live market depth (the top five levels on both sides) and the daily bhavcopy archive on nseindia.com. NSE Indices at niftyindices.com has the methodology PDFs, the factsheets that publish live constituent weights, the reconstitution announcements and the TRI series. Add NSE's daily FII/DII activity and the bulk- and block-deal lists, AMFI's monthly SIP and AUM data at amfiindia.com, and SEBI's own F&O participant studies on sebi.gov.in. For the US, the free LOBSTER sample files at lobsterdata.com are reconstructed limit order books built from Nasdaq's ITCH message feed. They're the cheapest way to watch a real book evolve message by message. The SEC's MIDAS market-structure datasets sit at sec.gov/marketstructure. S&P Dow Jones Indices at spglobal.com/spdji has the S&P 500 methodology PDF and factsheets, and Nasdaq's index methodology is at indexes.nasdaqomx.com. FINRA covers short interest. Finally, Jay Ritter's IPO data at site.warrington.ufl.edu/ritter gives you decades of free underpricing statistics. That panel turns the underpricing puzzle from an anecdote about Airbnb into a distribution.

# The book as a function of size, and the index as a weight vector — recomputed in-session (R3)
import numpy as np
ASKS = [(400,249.80),(600,249.90),(1000,250.00),(1500,250.20),(2500,250.50)]   # module's book
MID  = 249.65
def walk(levels, qty):                       # consume the book top-down; return avg fill + impact
    left = qty; cost = 0.0
    for q, p in levels:
        take = min(left, q); cost += take*p; left -= take
        if left <= 0: break
    done = qty - left
    return cost/done, 1e4*(cost/done - MID)/MID
for q in (100, 1500, 2500, 5000):
    print(q, *[f"{x:.4f}" for x in walk(ASKS, q)])   # 249.8000/6.0  249.9067/10.3  249.9840/13.4  250.1820/21.3
sh, px, iwf = np.array([100.,200.,50.]), np.array([500.,150.,800.]), np.array([.40,.75,.90])
ff = sh*iwf*px; w = ff/ff.sum()
print(np.round(100*w,2), np.round(100*(sh*px)/(sh*px).sum(),2), f"{1/(w**2).sum():.2f}")
# [25.48 28.66 45.86] float  vs  [41.67 25.00 33.33] full-cap  ->  2.80 effective names of 3

Verify. Prove both functions against the answers worked above. That's why you were given them. For the book, the spread is ₹0.30, the mid ₹249.65 and the spread 12.0 bps. Scenario A fills 1,500 shares at an average ₹249.9067 for 10.3 bps of impact. Scenario C's marketable limit fills 700 at ₹249.8429 and leaves 300 resting. Scenario D's market sell averages ₹249.465 for 7.4 bps.

For the index, check float weights of 25.48 / 28.66 / 45.86% against full-cap weights of 41.67 / 25.00 / 33.33%. You should get a divisor of 78.5, a day-two level of 1,009.04 and a return of +0.9045%. That return equals the float-weighted average of the three member returns to within 6 × 10⁻¹⁷. It's an identity worth checking, because it's what "an index return" actually means. Then comes a post-replacement divisor of 86.38 that leaves the level unchanged.

The applications come last. Index inclusion at a 1% weight against ₹4,00,000 crore of tracking assets is ₹4,000 crore, or 13.3 days of a ₹300 crore average daily traded value. The rights issue's TERP is ₹420 with each right worth ₹120. That leaves the subscriber and the right-seller both at ₹1,800 and the do-nothing holder at ₹1,680. The buyback retires 40m shares for +11.11% EPS accretion and lifts a 1m-share holder from 0.2500% to 0.2778% of the company. Every one of those is arithmetic your function must reproduce before you point it at a real book.

Quantitative lens

In Phase 0, the order book is the one place where a real quantitative object is already in front of you. So the findings below are about the book and index built above. They're properties of that book. None of them measures any real market.

Impact cost is a curve, it is convex, and the quoted spread is a floor you will never actually pay. On this book, an infinitesimal buy pays the half-spread, 6.0 bps. Fifteen hundred shares pay 10.3 bps. Twenty-five hundred pay 13.4 bps, and five thousand pay 21.3 bps. Doubling an order from 1,500 to 3,000 shares raises the cost per share from 10.3 to 14.8 bps, a factor of 1.44. So the total bill rises by a factor of 2.87, where you might have guessed 2.

Convexity is why institutions work orders over hours instead of sending them all at once. It's also why depth matters more than the spread alone. A tight spread tells you the price of the first share. Only the depth behind it tells you the price of yours.

The cost you cannot see is roughly the size of the entire cost you can. A 1,500-share round trip through this book pays 18.6 bps of impact, against the 23.3 bps all-in statutory bill that the toolkit computes line by line for a ₹3,00,000 delivery round trip: STT, stamp duty, exchange and SEBI fees, the DP charge and GST together. The two are the same order of magnitude, and only one of them appears on the contract note. Budget from the broker's tariff sheet and you've budgeted about half the transaction cost. The half you left out grows with position size and worsens in thin small-caps. Patience can reduce it, which makes it the half you actually get to decide on.

Free-float adjustment reorders an index; it does not fine-tune one. Moving the three-stock basket from full market cap to free float shifts 16.19 percentage points of index weight, nearly a sixth of the whole thing. It also inverts the size ranking, turning the largest company into the smallest weight. Remember the remark that a huge but 75%-promoter-owned Indian company can carry a modest index weight? It's no edge case. On any basket with dispersed promoter holdings the float step is one of the largest single adjustments in the construction.

A weighted index holds fewer names than it lists. The effective number of constituents is 1 / Σwᵢ², and the three-stock index above carries 2.80 of its 3. Apply the same arithmetic to the stated concentration figures, treating them as numbers to verify, and the ceiling is stark. Say the top ten names are 57.5% of a fifty-stock index. Take the most favourable case, with the top ten equal to each other and the remaining forty equal to each other. Even then it behaves like 26.6 equally weighted names. At 55% the ceiling is 28.3, and at 60% it's 25.0. For a five-hundred-stock index with 37.5% in its top ten the ceiling is 67. Both are ceilings, since any dispersion inside those groups pushes the real figure lower. So when "the index rose" describes the move of six stocks, nobody is exaggerating. A concentrated weight vector means that, arithmetically.

Now the limits, and they matter, because a market-microstructure number invites more confidence than it can carry. The book above is a snapshot. Real books refill between your fills, so a patient order often pays less than the walk predicts, and an urgent one in a falling market often pays much more. The walk also assumes nobody reacts to you. That's false for any order large enough to matter. Market makers widen and other participants trade ahead of visible size, and that reaction is often larger than the mechanical impact computed here. The convexity result belongs to this five-level book. A different depth profile gives a different curve, so recompute it on the depth you actually face. Don't just quote 1.44. The index results use a three-stock toy, and the concentration ceilings depend on figures you've been told to verify in the current factsheets. And none of it says anything about whether a price is right. All these numbers are about the cost of transacting, and that question is separate from what the thing is worth.

Do it in code. Write the depth walk as a function of size and plot the whole curve, instead of reading one number off it. Before you send an order, compute it as a fraction of the visible depth. Build the index as a weight vector and report 1/Σwᵢ² alongside the level. Then decompose any day's index move into per-stock contributions. It takes one line, and it cures you for good of treating an index as a single opinion.

Where this goes next: galaxy cross-links

• Execution algorithms and market microstructure (QD2.01). This is the professional version of the order book. The impact curve gets fitted instead of walked, and you'll meet VWAP and implementation-shortfall benchmarks. You'll also see the algorithms that exist because the curve above is convex.
• The payment system and plumbing (MS1.02). Novation, pay-in and pay-out and the depository legs, taken down to the settlement rails themselves. The same machinery that makes a T+1 equity trade final also clears everything else.
• Central-bank operations, reserves and the repo and money-market complex (MS1.03). The G-sec market is RBI turf and the ten-year yield is made there. Here that sentence becomes a market you can read, and you'll see where the risk-free rate of every later valuation comes from.
• pandas I, data wrangling (DA1.02). Bhavcopy files, factsheet constituent tables and FII/DII series are all just dataframes. Joining them stops being a chore and becomes the first thing you do before asking any question about flows.
• Failure modes, verification and the primary-source guardrail (AI0.06). The general treatment of the specific trap set here: a training cutoff, a rule that changed after it, and an answer delivered with complete confidence.

Teach it back & journal

Feynman prompt. Write a one-page explainer titled "Where does a stock's price come from?" for a smart 15-year-old. Cover what an order book is and why the price moves when someone buys aggressively. Explain why "the market decided" means nobody decided. Then add one plain sentence about what the price does not tell you about the business. Use no jargon beyond "bid" and "ask", and explain every other term in plain words. To test it, read it to someone. If they can then answer "who sets the price?" correctly, you pass.

Journal reflection. Before this week, what did you believe happened to your money when you "bought a share in a company"? Write the before-and-after in four sentences. IPO allocations, index inclusion and FII/DII flows all move prices without touching business fundamentals. Which of them would most tempt you to mistake flows for facts? And what one-line rule will you write into your future checklist to guard against it?

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