Event-Driven Special Situations and the Canon's Unclosed Drills

Phase 11 · Elective lab track · ~16 focused hours · Mastery gate >= 85%

A holder of 800 shares in a company that announces a rights issue at ₹400 while the stock trades at ₹640 has three choices on the record date, and one of them costs her ₹52,363.64. She can put up ₹1,20,000 and subscribe. She can sell the entitlement to somebody who will. Or she can leave the letter of offer unopened, which is what most retail holders do, and hand the difference to whoever subscribes in her place.

Nobody stole it. The arithmetic did it, and the arithmetic is on the first page of the offer document.

That transfer is the whole subject here. A corporate event splits a company's shareholders into people who must act and people who must not, and then it prices the security while the people who must not act are selling. An index fund receives a spin-off it cannot hold and sells it at any price on day one. A bond fund receives a warrant it has no mandate to own. A retail holder receives a rights entitlement and does nothing. In each case the seller is answering a question that has nothing to do with what the security is worth, and somebody on the other side is being paid to answer the question that does.

The elective that precedes this one taught the deal itself: how to model a merger, price an arbitrage spread and follow an Indian scheme of arrangement through the tribunal. The credit elective taught the waterfall and the covenant. What is left over is the awkward, unglamorous middle: the paper distributed in a deal that nobody wanted, the stub left standing after a recapitalisation, the entitlement that lapses, the estate that is worth more dead than alive. Joel Greenblatt catalogued these in You Can Be a Stock Market Genius (1997), and they have been named in passing here without ever being run. Six drills from the investing canon have been in the same position: Seth Klarman's liquidation and private-market anchors, Peter Lynch's category triggers and two-minute drill, Philip Fisher's don'ts, and Buffett's one-dollar test on a decade of retained earnings. They are drills, not doctrine, and a drill you have never run is a drill you do not have.

Two warnings, both meant. The first is that these situations are small. Their size is the reason they are mispriced and it is also the reason a professional cannot buy them, so anyone running institutional money should expect the opportunity set to be uninvestable at scale. The second is sharper. A leveraged stub, a rights entitlement in a promoter-controlled company and a fulcrum claim in an insolvency all share one property: the loss case is a total loss. The instruments are levered by construction, and the arithmetic that makes them attractive at 2% of a portfolio makes them ruinous at 20%.

Learning objectives

By the end you can:

1. Apply four tests to any announced corporate event and separate a genuine dislocation from a discount the security deserves: a named price-insensitive seller you can size in days of trading volume, a gap to a conservative value that survives the cost of carry, a dated mechanism that closes the gap, and an honest answer to what the seller knew.
2. Work a rights issue end to end: the theoretical ex-rights price, the value of the right per new and per existing share, and the wealth of a holder who subscribes, who renounces and who does nothing, with the three outcomes reconciling to the rupee.
3. Compute the promoter stake before and after a rights issue under full subscription, under partial public take-up with the shortfall absorbed, and under a scaled-down issue, and price the value that moves between the two groups.
4. Value the securities distributed in a deal: a low-coupon note priced off a comparable yield, and a warrant priced with a dilution-adjusted option model, then invert both market prices into the implied yield and the implied volatility that the forced sellers are accepting.
5. Build a leveraged recapitalisation and the stub it leaves behind: the special dividend, the post-recap leverage, the stub's share of enterprise value, its effective leverage, a five-outcome value ladder, and a position size that satisfies a stated ruin constraint rather than a Kelly fraction.
6. Compute a liquidation value asset by asset with defensible haircuts, wind-down costs and the full liability stack, then cross-check it against a private-market value from precedent transactions and a going-concern discounted value, and say which of the three anchors binds and which is the floor.
7. Classify a company into one of six categories and write the sell trigger the category demands, computing the trigger price where arithmetic supplies one: a stalwart's ceiling multiple, a fast grower's de-rating, a cyclical's mid-cycle multiple, a turnaround's justified price-to-book, and an asset play's discount.
8. Run Buffett's one-dollar test over a decade with the issuance and buyback adjustment shown, state the pass-fail reading, and quantify how much of the answer came from multiple re-rating and how much moves if the window shifts by one year.
9. Write a two-minute drill and a set of pre-registered sell criteria for one company, and score a governance gate that decides whether an Indian restructuring creates value for every holder or moves it to a controlling family.
10. Name the structural reason these mispricings persist, and say honestly which of them a person managing other people's money on a quarterly reporting cycle can and cannot act on.

Prerequisites & connections

Builds on. M1.03 for the equity build, rights and bonus issues and the reserves that fund them. M0.02 for the theoretical ex-rights price at its simplest, which is extended here into the promoter arithmetic and the renunciation decision. M3.06 for comparable companies and precedent transactions, which supply the private-market anchor. M3.10 for sum-of-the-parts, the Indian holding-company discount, distress and equity as an option; M3.10 keeps the spin-off and separation mechanics and this node cites rather than rebuilds them. E11.01 for the credit waterfall, covenants, recovery and the fulcrum security. E11.02 for deal anatomy, sources and uses, accretion and dilution, spin-offs and merger arbitrage. E11.04 for the promoter-group map, the related-party web and the disclosure ladder that the governance gate here depends on. M9.01 for position sizing and the ruin constraint that caps every stub in the practice set. M6.01 and M6.02 for the schools themselves, whose unfinished drills this node completes.

Feeds forward. E11.05: Market-Level Valuation, the Index Default and Your Own Policy is the sibling elective and the natural pair. That node decides how much of a portfolio has no named edge and therefore belongs in an index, and it writes the allocation band and the rebalancing rule that a policy needs; the drills here decide what the remaining sleeve is allowed to hold and at what size. M9.03 takes the category sell triggers written here into the full five-reason sell framework. M8.02 and M8.06 take the two-minute drill into the teardown and the document lab. M10.01 and M10.02 are where a special situation, if you find one worth the words, becomes a written thesis.

The deliberate non-overlaps run as follows. E11.02 owns spin-offs, demergers, carve-outs, merger arbitrage and the sources-and-uses table; a spin-off appears here only as a source of forced sellers to be sized, and the separation mechanics are cited, not repeated. E11.01 owns the credit waterfall, covenant analysis, rating methodology and the insolvency recovery ladder; the fulcrum security appears here as an equity opportunity and hands back to E11.01 the moment the question becomes a credit question. M3.10 owns sum-of-the-parts, the holding-company discount, real options and the distressed toolkit. E11.04 owns promoter mapping, related-party materiality, pledge mechanics and auditor signals; the governance gate here scores an event and cites that machinery. M6.01 and M6.02 own the schools and their biographies, and each drill below names the school it completes and hands the reader back to it. M3.07 owns the ten-year capital-allocation scorecard, in which the one-dollar test is one row of eight; the full mechanics of that row, including the adjustment nobody makes, live here. M0.03 owns the fee-drag identity and E11.05 owns the index default.

Core teaching content

4.3 Rights offerings II: the structural forced seller, and the India overlay

Calling a non-subscriber careless misses what is happening to her. She is a forced seller, in a specific and predictable way, and the forcing is structural.

A retail holder in a small position faces a fixed cost in attention: read the letter of offer, work out the entitlement, fund the application, meet the date. Below some position size the cost of doing that exceeds the value of the right, so rational inattention produces lapsed entitlements at exactly the sizes where they are most common. An institutional holder faces a different constraint: a fund at its cash limit, or a mandate that caps a single name, cannot put new money into a holding even when the right is worth taking, so it renounces into a thin market and takes whatever the market offers. A promoter or controlling family faces the reverse constraint. It usually can fund the entitlement, and it often has a reason to fund more than the entitlement.

The result is that in a deeply discounted rights issue, wealth moves in a predictable direction: from holders who cannot or do not act, to holders who can. The size of the transfer is the value of the lapsed rights, and it is computable in advance.

India makes this sharper for three reasons that an analyst has to know by name.

The first is the letter of offer, filed with the regulator and the exchanges, which carries the issue size, the ratio, the price, the record date, the rights entitlement credited to each demat account, and the objects of the issue. The objects clause is the document's most useful page and the least read: it states what the money is for, line by line, and the analysis in the previous section runs directly on it.

The second is renunciation. Indian rights entitlements are credited to demat accounts as a separate, tradable instrument for a defined window, and they trade on the exchange. That is a genuine improvement on the older physical regime, because it turns "act or lose" into "act or sell", and it means the value of a lapsed right is now visible as a market price rather than only as arithmetic. The catch is liquidity: rights entitlements often trade at a discount to their theoretical value in a thin window, so the non-subscriber who sells late recovers less than the arithmetic says. That discount is a measurable thing and worth measuring rather than assuming.

The third is promoter subscription and the treatment of the unsubscribed portion. A promoter may state in advance an intention to subscribe to its full entitlement, and may additionally apply for shares not taken up by the public. Where the promoter absorbs a large shortfall at a deeply discounted price, its stake rises, and the arithmetic of that rise is the single most useful calculation in an Indian rights issue.

Run it on the same company. The promoter holds 52.00%, which is 6.24 crore shares, entitling it to 2.34 crore new shares; the public's 5.76 crore shares entitle it to 2.16 crore. If everybody subscribes, the promoter holds 8.58 of 16.50 crore shares and its stake is unchanged at 52.00%, which is the point of a pro-rata issue.

Now suppose the public takes up only 40% of its entitlement and the promoter absorbs the rest. The promoter ends with 9.876 crore of 16.50 crore shares, a stake of 59.854545%, up 7.85 percentage points. It acquired 1.296 crore shares at ₹400 that are worth ₹574.55, a gain of ₹226.21 crore, and that figure is identical to the value the lapsing public holders gave up. The transfer is exact, and it happened without a single rule being broken.

Change one thing and the picture changes completely. If the promoter does not absorb the shortfall, the issue is scaled down to 3.204 crore new shares raising ₹1,281.60 crore, the ex-rights price becomes ₹589.42, and the promoter's stake rises to 56.432518% purely because the public did not take up its own entitlement. The stake still rises. The ₹226 crore of value does not move, because nobody bought the lapsed shares cheaply.

The distinction between those two cases is the whole governance question, and it is visible in the letter of offer before the record date. A promoter that pre-commits to its entitlement only, in a fairly priced issue, funding a stated use with a defensible return, is doing its job. A promoter that pre-commits to absorb the entire unsubscribed portion, in a deeply discounted issue, at a moment when the public is least able to fund it, is running a legal, disclosed and highly effective creeping acquisition. Both are lawful. They are not the same event, and the price of one right tells you what the difference is worth.

4.4 Merger securities: the paper that arrives unwanted

When a deal pays with something other than cash or ordinary shares, the target's holders receive an instrument they did not choose. It might be a senior note, a preferred share, a contingent value right or a warrant. Whatever it is, three things are true of it on day one: its holders are the wrong holders, it is small, and nobody publishes a value for it.

The wrong-holder problem is the mechanism from the first section in its purest form. An equity fund that owned the target now holds a five-year note. Its mandate does not permit credit. Its systems do not price it. Its analysts do not follow it. The fund sells, and the buyer sets the price.

Valuing the paper is ordinary work done in an unusual place, and the discipline is to value each leg on its own terms rather than to value "the package".

A distributed note is priced off the issuer's own comparable yield, not off its coupon. A five-year note with a face of $10.00 and a 6.00% coupon, issued by a company whose five-year debt yields 9.25%, is worth 0.60 × 3.864551 + 10 × 0.642529 = $8.744021, which is 87.44% of face. The 12.56% discount is not distress. It is the difference between the coupon the deal negotiators found convenient and the yield the market charges, and a holder who marks the note at face because "it is a par instrument" is overstating the consideration by $1.26 a share.

A distributed warrant is priced with an option model and then adjusted for the dilution it will cause. A four-year warrant struck at $30.00 on a $26.50 share, at 38% volatility, a 4.25% riskless rate and a 1.20% dividend yield, is worth $7.459519 before dilution. If the issuer has 120 million shares and issues 8 million warrants, exercise creates new shares, so the value belongs to a larger base and the warrant is worth 7.459519 × 120 ÷ 128 = $6.993299. Skipping the dilution adjustment overstates the warrant by 6.7%, which is a small error compared with the one most people make, which is treating the warrant as worthless because it is out of the money.

Put the legs together. A package of $18.00 cash plus that note plus that warrant is worth $33.737320. If the note trades at $7.60 and the warrant at $4.10, the market is paying $29.700000, a gap of $4.037320, or 25.65% of the two non-cash legs. Invert those two prices and the gap acquires a meaning: the note's market price implies a yield of 12.788345%, which is 354 basis points wide of the comparable issuer's own curve, and the warrant's price implies a volatility of 21.98% against a 38% assumption. The forced sellers are not making a credit judgement or a volatility judgement. They are making an inventory decision, and both implied numbers are what an inventory decision looks like when you translate it into the language of the instrument.

That translation is the skill. Anyone can say a security is cheap. Saying that the seller is accepting a 22% volatility on a four-year warrant tells you how cheap, in a unit that can be compared with the risk you are taking.

4.7 Restructurings beyond the clean spin-off

The separation mechanics belong to the valuation and deals work you have already done: the taxonomy of spin-off, carve-out, split-off and tracking stock; the sum-of-the-parts that measures the conglomerate discount; the entitlement ratio; and the Indian scheme of arrangement from board approval through tribunal sanction to the record date. What is left, and what changes the answer, is the family of separations that are not clean.

A partial spin-off distributes only part of a subsidiary and leaves the parent holding a stake. The retained stake is the analytical problem, because it sits inside the parent as an investment whose value the market frequently ignores and whose disposal overhangs the subsidiary's stock. The stake is a promise of future supply, and the parent's stated intention regarding it is a fact worth extracting from the scheme document.

A tracking stock creates a separate class of the parent's own equity whose dividend and reported results follow one division, without any legal separation at all, so nothing is ring-fenced. The division's assets remain the parent's assets and are available to the parent's creditors. The board owes its duties to the whole company, including when it allocates costs between the tracked division and the rest. Tracking stocks were fashionable for a decade and largely stopped being issued because the discount they were created to close does not close, for the excellent reason that nothing structural changed.

A reverse merger into a listed shell and a slump sale of a division to a related entity are the two shapes that most often disguise a transfer as a restructuring. In both, the question is not whether the mechanics work but whether the consideration is fair, and the arithmetic is a single subtraction: what did the buyer pay, and what would an unrelated buyer have paid.

Across all of these the analytical move is the same. Take the scheme document, find the valuation report and the swap or entitlement ratio, price the same assets independently from comparable transactions, and compute the difference. Where the difference is small, the restructuring is what it says it is. Where it is large, the next section applies.

4.8 The governance gate: the same event, two outcomes

An Indian restructuring can create value for every shareholder or move value from the listed company to a controlling family's private vehicle, and the mechanics look identical from the outside. The forensics work teaches how to map a promoter group, quantify a related-party web and read the disclosure ladder. What the event analyst needs is a gate: a small set of checks that decide, before the record date, which of the two events this is.

Start with the arithmetic, because it fixes the size of what is at stake. Suryagarh Industries has 30.00 crore shares at ₹308, a promoter holding of 54%, and a consumer business that precedent transactions value at ₹4,100 crore. If that business is sold to a promoter-owned unlisted entity for ₹2,400 crore, the value gap is ₹1,700 crore. The promoter already owns 54% of that gap through its listed shareholding, so what it gains by moving the asset is the other 46%, which is ₹782 crore, or ₹56.67 for every one of the 13.80 crore public shares. Against a ₹308 share price that is 18.40% of a public shareholder's holding, moving in a single board resolution.

Compare the alternative. A clean demerger of the same business to all shareholders in the same proportion leaves a sum-of-the-parts of (4,100 + 7,900) − 1,150 = ₹10,850 crore of equity, or ₹361.67 a share, and the promoter's economic interest is unchanged at 54% of both entities. The market at ₹308 is pricing something close to the transfer version, at ₹305 a share, which is itself a piece of information about what the market expects.

Five checks separate the two, and they are available before the event completes.

• Who receives the asset. A distribution to all shareholders in their existing proportion cannot transfer value between them. Any structure in which the asset lands somewhere the public does not own pro rata is a transfer until proved otherwise.
• What the independent valuation says, and who chose the valuer. A registered valuer's report supports the ratio or the consideration. Read its methodology, its comparable set and its date, and compare its conclusion with your own precedent-transaction work. A valuation that uses book value for an asset the market prices on earnings is the most common and most effective device.
• Whether the approval required a majority of the minority. Where a transaction with a related party crosses the materiality gate, no related party may vote on the resolution, whether or not it is a party to that particular transaction, which converts the arithmetic above from a grievance into a veto. The gate is set on consolidated turnover in bands, 10% up to ₹20,000 crore of turnover and a smaller share of the excess above that, capped at ₹5,000 crore, and the bands move with the listing regulations, so read the current version before you apply it rather than the number in your memory (SEBI LODR Fifth Amendment 2025, as of October 2026; check against the listing regulations as amended). Suryagarh's turnover of ₹9,600 crore sits in the first band, so the gate is ₹960 crore, and the ₹2,400 crore consideration is 2.5 times that, so the vote is required. With 62% of public shares voting, 4.278 crore shares, or 14.26% of the register, can defeat it.
• What the audit committee minuted, and what it did not. The approval record is disclosed. A committee that recorded the alternative of an open auction and rejected it for stated reasons is doing something different from one that approved a valuation it received on the morning of the meeting.
• What the promoter's other entities did in the twelve months before. Loans, guarantees, share purchases and pledge changes in the promoter group are the context that makes a transaction legible. An asset moving to a vehicle that took on debt to buy it, six months after that vehicle was capitalised, is a story with more than one chapter.

None of these five is a moral judgement, and none requires an allegation. Each is a document, a number, or a vote count, and together they turn "this looks like a promoter-friendly scheme" into a computation with a number attached.

4.9 Klarman's three anchors, and which one binds

Seth Klarman's Margin of Safety (1991) is the most useful book ever written about downside, and its central practical contribution is that a business does not have one value; it has several, computed by different methods for different purposes, and the discipline is knowing which one is doing the work.

Three anchors, in our compression of his framework. The net present value of the cash flows the business will generate as a going concern, which is the standard discounted valuation. The liquidation value, which is what the assets fetch if the business stops. And the private-market or acquisition value, which is what an informed industry buyer would pay for the whole thing, read off precedent transactions rather than off trading multiples. Klarman also discusses stock-market value, meaning what comparable public companies trade at, and treats it with the least confidence of the four, for the sound reason that it inherits the market's own error.

The reason to compute all three rather than the best one is that their relationship carries information that no single number does.

When the going-concern value is highest, the business earns more than its cost of capital and its assets are worth more inside it than outside. Ordinary valuation applies and the other two anchors are only a floor.

When the private-market value is highest, the business is worth more to somebody else than to its current owners. That gap is either an acquisition case or a management indictment, and reading the difference is the skill.

When the liquidation value is highest, the business is destroying value by continuing to operate: the assets earn less deployed than they would fetch sold. That is Graham's net-net territory in modern clothes, and it comes with a warning attached, which is that a company earning below its cost of capital usually goes on earning below it for as long as the people running it are allowed to, and the liquidation value melts while you wait.

The floor is always the liquidation value, and it is a floor only in the specific sense that it bounds the loss if somebody chooses to realise it. In a controlled company where nobody will, it is a number, not a protection.

4.11 The performance derby, named

The mispricings above have been documented for decades and they persist. Any honest treatment has to explain why, and the explanation is not that professional investors are stupid.

Klarman's term for it is the performance derby, and the argument is structural. Institutional money is measured against a benchmark over short windows. A manager who deviates from the benchmark and is right is rewarded modestly; a manager who deviates and is wrong for four quarters loses assets and sometimes a job. The asymmetry is not in the returns, it is in the career, and it produces a set of entirely rational behaviours that together keep the small, ugly, complex opportunity uncontested.

Four consequences, ranked by how much of the effect they explain.

The largest is the tracking-error budget. A mandate that permits three percentage points of annual tracking error effectively forbids a portfolio from looking different from its benchmark, and every situation in this node looks different from every benchmark by construction. The constraint binds before any judgement about the security is made.

Second is the reporting window. A special situation with a dated catalyst nine months out will look like a mistake for two quarters, and the manager will be asked about it in both. This is why catalyst-driven investing is more common in vehicles with locked capital than in daily-dealing funds, and it is a structural fact rather than a failure of nerve.

Third is asset gathering. A fund that has grown to a size where a $960 million spin-off cannot move the needle stops looking at $960 million spin-offs, and the incentive to grow is stronger than the incentive to keep the opportunity set.

Fourth, and smallest but not zero, is the explanation cost. Owning a stub, a warrant or a rights entitlement requires explaining it to a committee that would rather not learn what it is.

Two honest qualifications belong with this argument. The first is that it explains persistence, not size: the derby explains why nobody competes away the opportunity, and it says nothing about whether the opportunity is large enough to be worth your time after costs and taxes. The second is that the same argument cuts against you. If you are managing other people's money on a quarterly cycle, you are inside the derby, and the correct conclusion may be that you cannot run these strategies rather than that you should. The individual investor's advantage here is real and it is one of the few places where it is genuinely structural rather than rhetorical.

4.13 The two-minute drill and pre-registered sell criteria

Lynch's second unfinished drill is the discipline of being able to say, out loud, in about two minutes and without notes, why you own a company. It sounds like a communication exercise. It is a diagnostic, because the constraint that makes it work is that you have to do it from memory, and anything you cannot hold in your head is something you have not understood.

Five parts, and they have to be in this order because each one constrains the next.

Start with the story: what the company sells, to whom, and why they buy it rather than the alternative. One or two sentences, no jargon, and no adjectives that a competitor could not also claim.

Then the reasons to own it now, meaning what is true today that makes the price attractive. The category from the previous section belongs in this part, because a fast grower's reason and a cyclical's reason are different kinds of statement and they are not interchangeable.

Third comes what has to go right: the two or three things that must happen for the thesis to work, stated as observable, dated events rather than as hopes. "Margins recover" is not one of them. "The new plant reaches 70% utilisation by the fourth quarter, taking the blended margin above 13%" is.

Fourth is what would break it, meaning the specific, checkable developments that would make you wrong. This is the hardest of the five and the one that turns the drill into pre-registered sell criteria, because a break condition written in advance is a sell order with a date already on it.

Last is what would make it a mistake rather than bad luck, and the distinction is the one the whole selling framework is built on. A thesis that broke for a reason you could not have known is bad luck. A thesis that broke for a reason that was on page 40 of the annual report is a mistake, and writing down which is which, before you know the answer, is how a decision journal becomes useful rather than decorative.

Run this on a company and one of two things happens. Either the five parts come out clean in two minutes, in which case you own the thesis and the sell criteria write themselves. Or you find yourself reaching for the model, in which case the honest reading is that you own a spreadsheet and not a company.

4.14 Fisher's don'ts, translated

Philip Fisher's Common Stocks and Uncommon Profits (1958) is usually taught for the fifteen points and the scuttlebutt method. Its two chapters of "don'ts" almost never are, and they are the most durable part of the book because errors survive better than techniques. Ten rules follow, in our own phrasing and ranked by how much money they save a learner, not by Fisher's order.

The one that costs the most, by a distance, is overpaying for a great business because the business is great. Fisher was the original quality investor and he still insisted that a superb company bought at the wrong price is a bad investment for years. The modern version is the compounder bought at 60 times earnings on the argument that quality is worth any price; the arithmetic in the stalwart trigger above is the antidote.

Second is abandoning a correct analysis because the price moved against you. Fisher's point is that a share price falling after you buy is information about the market's mood and not about the business, unless something in your five-part story has actually changed.

Third is losing a good position over a trivial difference in the entry price. An investor who has done the work, concluded that a business is worth ₹800, and then refuses to pay ₹412 instead of ₹408 has spent weeks of analysis to save half a percent and frequently misses the position entirely.

Fourth is diversifying past the point where you know what you own. Fisher argued that the standard prescription of very wide diversification was a confession of ignorance, and that a portfolio of twenty businesses you understand beats one of eighty you do not. The counterweight, taken seriously throughout, is the ruin constraint: concentration is defensible only when no single position can end you.

Fifth is buying a company because you like its annual report. A well-designed report is evidence about the investor-relations budget.

Sixth is treating a stock's own price history as information about its value. That a share has halved tells you nothing about whether it is cheap, and that it has doubled tells you nothing about whether it is expensive.

Seventh is selling a genuinely superior business because the market or the economy looks frightening. Fisher's version was about war scares. The modern version is an election, a rate cycle or a headline about a recession, and the response is the same: if the business will still be earning its returns in five years, a macro fear is a buying window and not a thesis.

Eighth is buying into an unproven promotional company on the strength of the story. Fisher's target was the newly floated company with no operating record; the shape recurs in every cycle and every market with a new label attached.

Ninth is letting an unwillingness to look at an unfamiliar listing venue or a small size shrink the opportunity set. This is the mirror of the first section's argument: small and awkward is where the mispricing lives.

Tenth, and the one Fisher would have put first, is following the crowd. Not as a general injunction to be contrarian, which is its own trap, but specifically: the accepted appraisal of an industry can be wrong for years, and the gap between what a business will earn and what the consensus thinks it will earn is the only durable source of return.

The one that most often costs a learner money is the first, and the reason is that it is the only one on the list that feels like discipline while you are committing it.

Common mistakes & how experts think differently

1. Reading the rights-issue discount as a gift. The 37.50% "discount" in the worked example is entirely erased by the ex-rights adjustment; the real question is whether the ₹1,800 crore earns the cost of equity. An expert reads the objects clause of the letter of offer before reading the price, because the price is arithmetic and the objects clause is the investment.
2. Treating a lapsed entitlement as a passive choice. Doing nothing is the only one of the three routes that costs money, and it costs a precise, computable amount. Experts diarise the record date and the renunciation window the day the letter of offer is filed, and treat "I did not get round to it" as a realised loss rather than an administrative slip.
3. Marking distributed paper at face. A 6.00% note from an issuer whose curve is at 9.25% is worth 87.44% of face on day one, and a package valued at face overstates the deal consideration. Experts value each leg on its own terms and then reconcile the total to the announced headline value, which is almost always a different number.
4. Calling a warrant worthless because it is out of the money. The warrant in the worked example is $3.50 out of the money and worth $6.99 before dilution, because it has four years and 38% volatility. Experts price the option and then adjust for the dilution the exercise will cause, and they notice that skipping the dilution step overstates the value by 6.7%.
5. Sizing a stub with a Kelly fraction. Kelly on the worked distribution says 23.77% of capital and half-Kelly says 11.88%, against a 15% chance of total loss. Experts apply the ruin constraint first and the growth-optimal calculation second, because Kelly assumes an infinite sequence of bets and an investor who cannot be fired.
6. Confusing the two waterfalls. The payment-priority waterfall inside a securitisation and the statutory distribution ladder in an insolvency are different objects that share a name. Experts can say which one they are in before they rank a claim, and they know that the fulcrum security in India is usually a bank loan that does not trade on an exchange.
7. Taking a valuation report as independent because it is signed. A registered valuer's report supports the consideration in a related-party transfer, and its methodology, comparable set and date are the parts that decide whether it is evidence. Experts reprice the same asset from precedent transactions and treat the difference as the number under dispute.
8. Quoting a liquidation value as a point estimate. The worked example's ₹126.15 becomes ₹48.34 or ₹191.69 under defensible alternative haircuts on land and plant. Experts publish the band and name the two assumptions that produce it, because a floor that moves by a factor of four is not a floor until you say how wide it is.
9. Buying the old equity in an insolvency because it is cheap. A share priced at ₹3 in a company whose resolution plan will extinguish it is not cheap, it is priced. Experts identify the impaired claim that becomes the new equity and buy that, or they conclude that the trade is unavailable in a public market and pass.
10. Running the one-dollar test without the buyback adjustment. The raw comparison gives −0.93 where the adjusted one gives 0.20, and in the passing example it understates the ratio instead. Experts compute the adjusted version, then decompose it to show how much came from the multiple, and only then form a view about management.
11. Calling a category without writing its trigger. Labelling a company a fast grower is worth nothing on its own; the value is in the sentence that says what observable number, at what level, for how long, ends the thesis. Experts write the trigger on the day they buy, with a date, and treat a trigger written after a price move as a rationalisation.
12. Assuming the performance derby argument is a licence rather than a constraint. The structural reason these opportunities persist applies to you if you manage other people's money on a quarterly cycle. Experts state honestly which strategies their own capital structure permits, and decline the rest.

Worked examples

Every figure below was recomputed in Python before it was written down. Companies are synthetic and labelled so, except in the one-dollar test, where one real decade is used and its inputs are rounded public-record figures carrying a verify instruction and a date. Currencies are marked in the heading.

Worked example 1: A rights issue, from the ex-rights price to the promoter's stake (India, ₹)

Kaveri Speciality Chemicals Ltd (synthetic) has 12.00 crore shares trading at ₹640. It announces a rights issue of three new shares for every eight held, at ₹400, to fund a capacity expansion. The promoter group holds 52.00%. Compute the ex-rights price, the value of the right, the position of a holder of 800 shares under each of her three choices, and the promoter's stake after the issue under full and partial subscription.

Solution. Step 1, the issue. New shares = 12.00 × 3/8 = 4.500000 crore. Cash raised = 4.50 × 400 = ₹1,800.000000 crore. Shares after = 16.500000 crore. Market capitalisation before = 12.00 × 640 = ₹7,680 crore.

Step 2, the theoretical ex-rights price. (7,680 + 1,800) ÷ 16.50 = ₹574.545455.

Step 3, the right, two ways. Per new share: 574.545455 − 400 = ₹174.545455. Per existing share: (640 − 400) × (3/8) ÷ (1 + 3/8) = ₹65.454545, which equals 640 − 574.545455. The two routes agree, which is the check that the share counts are right.

Step 4, the headline against the reality. The advertised discount is (640 − 400) ÷ 640 = 37.500000%. The discount to the price the share will actually trade at is (574.545455 − 400) ÷ 574.545455 = 30.379747%.

Step 5, the holder of 800 shares. Entitlement = 800 × 3/8 = 300 new shares, costing ₹1,20,000. Cum-rights wealth = 800 × 640 = ₹5,12,000.

The cost of inaction is ₹52,363.64, which is 800 × ₹65.454545.

Step 6, the promoter. Promoter shares = 12.00 × 0.52 = 6.240000 crore, entitlement 2.340000 crore. Public shares 5.760000 crore, entitlement 2.160000 crore.

In Case B the promoter's stake rises 7.854545 percentage points and it acquires 1.296000 crore shares at ₹400 that are worth ₹574.545455, a gain of 1.296 × 174.545455 = ₹226.210909 crore. That figure is identical to the value the lapsing public holders gave up, which is the test that the transfer has been computed correctly. In Case C the promoter's stake still rises, to 56.432518%, and the ex-rights price becomes (7,680 + 1,281.60) ÷ 15.204 = ₹589.423836, but no value changes hands because nobody bought the lapsed shares.

Step 7, the use of proceeds. At a 13% cost of equity, ₹1,800 crore deployed at 8% is worth 1,800 × 0.08 ÷ 0.13 = ₹1,107.692308 crore, destroying ₹692.307692 crore, or ₹41.958042 a share, which is 7.302824% of the ex-rights price.

Read it. Steps 1 to 5 are arithmetic and cannot be argued with; step 6 is the governance question and step 7 is the investment. A holder who subscribes correctly and ignores step 7 has protected herself from a ₹65 transfer and walked into a ₹42 one. The number to take away is that the case for subscribing is never the discount, and the ₹226 crore in Case B is a legal, disclosed transfer that no rule prevents.

Worked example 2: A leveraged recapitalisation, the stub, and how large a position it justifies (US, $)

Ridgeline Components Inc (synthetic) has 40.0 million shares at $25.00 and $200 million of net debt, on EBITDA of $150 million. It borrows $700 million and pays the proceeds out as a special dividend. Compute the stub, its effective leverage, its value across five enterprise-value outcomes, and the position size a 2% ruin constraint permits when the stub trades at $6.60.

Solution. Step 1, before. Equity 40.0 × 25.00 = $1,000m; enterprise value 1,000 + 200 = $1,200m; EV/EBITDA = 8.000000×; net debt/EBITDA = 1.333333×.

Step 2, the recapitalisation. Special dividend 700 ÷ 40.0 = $17.500000 a share. Net debt after $900m, or 6.000000× EBITDA.

Step 3, the stub under conservation of value. 1,200 − 900 = $300m, which is $7.500000 a share, and 25.00 − 17.50 = 7.50 confirms it.

Step 4, the tax shield. At 24%, 700 × 0.24 = $168m, lifting enterprise value to $1,368m and the stub to $11.700000 a share. The honest model is the range $7.50 to $11.70.

Step 5, the leverage. Stub equity is 300 ÷ 1,200 = 25.000000% of enterprise value, so its effective leverage is 4.000000×.

Step 6, the ladder.

Step 7, the market's view. At $6.60 the stub is 12.000000% below the no-shield value and 43.589744% below the full-shield value. The market is paying nothing for the tax shield and discounting the equity beyond that, which is what an orphaned security looks like after a large special dividend.

Step 8, expected value and size. With probabilities of 0.15, 0.20, 0.30, 0.25 and 0.10 the expected stub value is $7.625000, an expected return of +15.530303% from $6.60. Full Kelly on that distribution is 23.769899% of capital and half-Kelly 11.884949%. A ruin constraint that caps the loss from any one position at 2% of capital, against a worst case of a total loss, permits 2.000000%. Half-Kelly is 5.942475 times larger than the ruin cap.

Read it. The stub is a 4-times-levered claim whose expected return is genuinely attractive and whose loss case is total. Both facts are in the same table, and only one of them survives being sized wrongly. The tax-shield range is the honest part of the answer: anyone quoting a single fair value for a stub has hidden an assumption about how permanent the new leverage is, and the $4.20 spread between $7.50 and $11.70 is how much that assumption is worth.

Worked example 3: Valuing the securities distributed in a deal (US, $)

Calder Instruments Corp (synthetic) is acquired by Northwind Group (synthetic). Each Calder share receives $18.00 in cash, one Northwind senior note with a face of $10.00 paying 6.00% annually and maturing in five years, and one warrant to buy one Northwind share at $30.00, expiring in four years. Northwind's five-year debt yields 9.25%. Its shares trade at $26.50 with a 1.20% dividend yield and 38% volatility; the riskless rate is 4.25%; it has 120 million shares outstanding and is issuing 8 million warrants. The note trades at $7.60 and the warrant at $4.10. Value the package and say what the market prices imply.

Solution. Step 1, the note. Five-year annuity factor at 9.25% = 3.864551; five-year discount factor = 0.642529. PV = 0.60 × 3.864551 + 10 × 0.642529 = $8.744021, which is 87.440210% of face.

Step 2, the warrant before dilution. d1 = 0.377299, d2 = −0.382701, N(d1) = 0.647024, N(d2) = 0.350971. Call value = 26.50 × e^(−0.012×4) × 0.647024 − 30 × e^(−0.0425×4) × 0.350971 = $7.459519.

Step 3, the dilution adjustment. 120 ÷ (120 + 8) = 0.937500, so the warrant is worth 7.459519 × 0.9375 = $6.993299. Omitting this step overstates the warrant by 6.67%.

Step 3a, the independent check. Price the same warrant on a Cox-Ross-Rubinstein binomial tree instead of the closed form. At 4,000 steps it returns $7.459512 undiluted and $6.993292 diluted, against the closed form's $7.459519 and $6.993299, a difference of seven parts in ten million. Two routes to the same number is the standard for an option value, because a mis-signed rate or a mis-scaled volatility shows up as a large disagreement and not as a small one.

Step 4, the package.

The gap is $4.037320 a share, which is 25.654432% of the two non-cash legs.

Step 5, invert the market prices. Solving for the yield that prices the note at $7.60 gives 12.788345%, which is 353.83 bp wide of the issuer's own five-year curve. Solving for the volatility that prices the warrant at $4.10 gives 21.982545%, against the 38% assumption.

Read it. The point of step 5 is that neither implied number is a considered judgement about Northwind's credit or its volatility. A 354 basis-point premium over the same issuer's own comparable debt cannot be a credit view, because it is the same issuer. What it is, is the price at which an equity fund with no mandate to hold credit will clear its inventory. Two cautions belong with the answer. The 38% volatility is an assumption and the warrant value is roughly linear in it over this range, so a reader who thinks 30% is right gets a materially smaller number. And both market prices are only quotes until you have checked that the instrument actually trades. A wide indicative quote on a security that has not traded in a week is not a price.

Worked example 4: Three anchors on an asset-heavy company (India, ₹)

Konkan Rolling Mills Ltd (synthetic) has 2.40 crore shares trading at ₹105. It is an asset-heavy steel processor earning below its cost of capital. Compute its liquidation value asset by asset, its private-market value from precedent transactions, and its going-concern value, and say which anchor binds.

Solution. Step 1, the estate.

Step 2, the cost of stopping. Nine months of running cost at ₹14 crore a month = ₹126.00 crore. Severance of six months' payroll at ₹9 crore a month = ₹54.00 crore. Professional and process fees at 3.5% of realisations = ₹60.851000 crore. Total ₹240.851000 crore, which is 13.85% of gross realisation. Net realisation ₹1,497.749000 crore.

Step 3, the claims. Secured term debt ₹640, unsecured debt ₹180, trade payables ₹305, employee dues ₹48, statutory dues ₹22, a total of ₹1,195.00 crore. Residual to equity ₹302.749000 crore, or ₹126.145417 a share.

Step 4, the band. Re-run step 1 with land at a 30% haircut and plant at 85%: ₹48.342292 a share. Re-run it with land at 5% and plant at 55%: ₹191.685000. The band is ₹143.34 wide, which is larger than the central estimate.

Step 5, private-market value. Normalised EBITDA ₹230 crore at the 6.5 times paid in recent transactions for comparable assets gives an enterprise value of ₹1,495.00 crore; net debt is 640 + 180 − 85 = ₹735.00 crore; equity ₹760.00 crore, or ₹316.666667 a share.

Step 6, going concern. Normalised free cash flow ₹95 crore growing at 3% discounted at 13.5% gives an enterprise value of ₹931.904762 crore, equity of ₹196.904762 crore, and ₹82.043651 a share.

Read it. The ordering is the finding. The going-concern value is the lowest of the three because the assets earn 95 ÷ 1,497.75 = 6.34% on what they would fetch, against a 13.5% cost of capital, so continuing to operate destroys value. The liquidation value is above the market price, which is the Graham case. The private-market value is far above both, which says an industry buyer would pay for the assets and run them differently. Those are three different theses and only one of them can be yours. The honest caution is that all three depend on somebody deciding to act. In a company where the board will neither sell nor liquidate, the ₹126 floor is a number on your spreadsheet and not a protection on your capital, and the ₹143 band on the estimate is wide enough that the floor itself is a range.

Worked example 5: The one-dollar test over a decade (US, $)

Run Buffett's retained-earnings test on two decades: a real one that failed and a synthetic one that passed, with the issuance and buyback adjustment shown and both caveats quantified.

Solution. Step 1, the failing decade. IBM, 2010 to 2019. Cumulative net income ≈ $125bn; cumulative dividends ≈ $50bn; cumulative buybacks ≈ $88bn; equity issued ≈ $3bn; market capitalisation from ≈ $190bn to ≈ $120bn. (Rounded to teaching scale from the published record and checked in September 2026; check against the 10-K filings before quoting. The capital-allocation work you have already done grades the same decade on eight criteria, and this is the mechanics of one of them.)

retained            = 125 - 50                    = $75bn
raw value created   = 120 - 190                   = -$70bn      ratio -0.933333
adjusted created    = (120 + 88) - (190 + 3)      =  $15bn      ratio  0.200000

Each dollar retained produced 20.00 cents of market value, and $60bn of shareholder capital was consumed. Flexing every input 10% in the company's favour gives 0.958261, still below the bar; flexing every input 10% against gives −0.271351. The verdict is robust to the rounding.

Step 2, the passing decade. Harrow Precision Corp (synthetic): net income $9.20bn, dividends $2.30bn, buybacks $3.40bn, equity issued $0.25bn, market capitalisation $6.10bn to $22.40bn.

retained            = 9.20 - 2.30                 = $6.90bn
adjusted created    = (22.40 + 3.40) - (6.10 + 0.25) = $19.45bn  ratio 2.818841

Step 3, caveat one, the multiple. Harrow's earnings went from $0.50bn to $1.40bn, and its multiple from 6.10 ÷ 0.50 = 12.200000× to 22.40 ÷ 1.40 = 16.000000×. At a constant 12.2 times the ending market capitalisation would be $17.080000bn, the adjusted value created $14.130000bn, and the ratio 2.047826. So 0.771014 of the ratio, which is 27.352185% of it, came from re-rating rather than allocation.

Step 4, caveat two, the window. End the test at year nine, with a market capitalisation of $14.60bn after a drawdown, $3.10bn of cumulative buybacks and $6.10bn retained: adjusted value created $11.350000bn and a ratio of 1.860656. One year of window moved the answer by 0.958185, or 33.99% of the ten-year reading.

Read it. The unadjusted ratio for the failing decade is −0.93 and the adjusted one is 0.20; both fail, but only one of them is a number you could defend in a meeting, and an analyst who quotes the raw figure will be corrected by anyone who knows what a buyback does. The two caveats cut in opposite directions and both need stating: re-rating gave Harrow more than a quarter of its score, and a one-year change of window would have cost it a third. A test this sensitive to its endpoints is a screen and a conversation-starter, not a verdict, and it is most useful when the answer is far from 1.0 in either direction.

Worked example 6: Six categories, ten companies, ten triggers (India ₹ and US $)

Classify ten companies into Lynch's categories and write each one's sell trigger as a number. The valuation inputs below are supplied for the drill rather than reported from the market. They are illustrative, they were chosen in September 2026 to make the arithmetic instructive, and the arithmetic rather than the inputs is what is being learned. Company names fix the category only, which is a structural judgement that changes slowly; pull the current multiple, growth rate and yield before you write a real trigger.

Solution.

Rows 2 and 9 are the drill's whole point. The commercial-vehicle maker shows a ratio of 0.2375, which on a naive reading is the cheapest name on the page; it looks cheap because 40% growth off a cyclical trough measures a recovery rather than a growth rate, so the ratio is meaningless. The aluminium producer at 7.0 times has no usable ratio at all.

The triggers, computed:

Check the triggers against each other. Each trigger price has to reconcile with the multiple it came from: 19.80 ÷ 50.0 − 1 = −60.400000%, 26.00 ÷ 52.0 − 1 = −50.000000%, 12 × 1.60 = 19.20 and 19.20 ÷ 29.40 − 1 = −34.693878%, and 5,600 ÷ 2,050 − 1 = +173.170732%. If a trigger price does not reproduce its own percentage, the growth rate and the multiple have been taken from different periods, which is the most common error in the whole table.

Read it. Every one of those triggers is arithmetic you can do before you buy, and every one of them tells you something uncomfortable. The stalwart's trigger sits 60% below its price, which means the conservative-sounding rule you adopted would have you sell at a price you never expect to see, so either the rule or the purchase is wrong. The turnaround is priced as though the fix has already worked. The cyclical looks cheapest exactly where it is most dangerous. The honest limit of this drill is that the categories are judgements and a company can move between them. The usual path runs from fast grower to stalwart, and then to slow grower a decade later, and a trigger written for the old category is worse than none.

Worked example 7: Sizing a forced seller, and scoring the four tests (US, $)

Meridian Aerospace Holdings (synthetic), market capitalisation $18.0bn on 300 million shares, spins off Vantage Sensors (synthetic) at one Vantage share for every four Meridian shares. Index funds and index-constrained mandates hold 41% of Meridian. Vantage trades at $12.80 when-issued, has EBITDA of $148 million, carries $180 million of net debt at 7.2%, and peers trade at 9.0 times EBITDA. Expected daily traded value is 0.9% of market capitalisation. Score the four tests.

Solution. Test 1, the mechanism. Vantage shares issued = 300 × 0.25 = 75.000000 million; market capitalisation 75 × 12.80 = $960.000000m. Forced holders own 0.41 × 75 = 30.750000 million shares worth $393.600000m. Daily traded value 0.009 × 960 = $8.640000m. Days of volume held by forced sellers = 393.60 ÷ 8.64 = 45.555556. A named seller holding 46 days of volume scores 2.

Test 2, the gap. Vantage enterprise value at peer multiples 148 × 9.0 = $1,332m; equity 1,332 − 180 = $1,152m; fair value 1,152 ÷ 75 = $15.360000 a share. The when-issued price is 16.666667% below that, an upside of +20.000000%. Over an assumed seven months that annualises to +36.690798%, and at a 20% capital-gains rate the after-tax return on the gap is +16.000000%. Seven months does not reach the long-term holding period in the United States, so 20% is the optimistic end of the tax assumption and an ordinary-income rate is the realistic one. A gap that survives tax and time scores 2.

Test 3, the closer. Index reconstitution at the next scheduled review, the first standalone quarterly result, and the first analyst initiation are all dated but none is contractual, and the reconstitution date is the only one that reliably ends the selling. Scores 1.

Test 4, what the seller knew. Vantage carries 180 ÷ 148 = 1.216216× net leverage with interest of $12.960000m and EBITDA cover of 11.419753×, which is comfortable, so the parent did not dump its debt here. The unresolved question is why the parent chose to separate a business with these economics, and the scheme documents give a strategic rationale rather than a financial one. Scores 1, because the question is answered adequately rather than fully.

The total is 6 of 8 with no zero, so the situation passes.

Read it. Test 1 is the one that turns a story into a number, and 45.6 days is the whole argument: the selling has a size and an end. Two honest limits. The 0.9% daily-volume assumption is doing a great deal of work, and at 1.8% the overhang halves to 22.8 days, so the assumption belongs in the write-up next to the answer. And the 9.0 times peer multiple is the second assumption; at 8.0 times the fair value is $13.386667 and the gap collapses from 20.00% to 4.58%, which would not survive the tax and the carry. Both sensitivities should be computed before the position is taken, not after.

Worked example 8: The governance gate on an Indian restructuring (India, ₹)

Suryagarh Industries Ltd (synthetic) has 30.00 crore shares at ₹308, a promoter holding of 54%, net debt of ₹1,150 crore, and two businesses. Precedent transactions value the consumer business at ₹4,100 crore and the remaining business at ₹7,900 crore. Turnover is ₹9,600 crore. The board proposes to sell the consumer business to an unlisted promoter-owned entity for ₹2,400 crore. Quantify what is at stake and score the gate.

Solution. Step 1, the value gap. 4,100 − 2,400 = ₹1,700 crore.

Step 2, who loses it. The promoter already owns 54% of the listed company and therefore 54% of the gap. What it gains by moving the asset into a wholly owned vehicle is the public's 46%: 0.46 × 1,700 = ₹782.000000 crore. Public shares number 30.00 × 0.46 = 13.800000 crore, so the loss is 782 ÷ 13.80 = ₹56.666667 a share, which is 18.398268% of the ₹308 price.

Step 3, the alternative. A pro-rata demerger to all shareholders gives a sum of the parts of (4,100 + 7,900) − 1,150 = ₹10,850 crore of equity, or ₹361.666667 a share, +17.424242% above the market. The promoter's economic interest is unchanged at 54% of both entities.

Step 4, what the market is pricing. After the transfer the listed equity is worth (7,900 + 2,400) − 1,150 = ₹9,150 crore, or ₹305.000000 a share. The market at ₹308 is within 1% of that, which says it already expects the transfer version.

Step 5, the gate.

Read it. The arithmetic converts a suspicion into ₹56.67 a share, and the majority-of-minority arithmetic converts the grievance into a vote that 14.26% of the register can win. Neither of those is an allegation about anybody's conduct, and both are computable from public documents before the record date. The honest limit is that ₹4,100 crore is an estimate from comparable transactions, and if the right figure were ₹3,000 crore the gap would be ₹600 crore and ₹20.00 a share. The precedent set, and its dates, is where an argument about this transaction would actually be won or lost.

Practice set

Work each problem before reading the solution. A calculator is assumed and a spreadsheet is better. Answers within 0.5% of the stated figure count as correct except where a problem asks for a percentage or a percentage-point answer, where the band is 0.1 percentage points. Where a problem asks for a judgement as well as a number, write the judgement in one sentence before you check it, because that sentence is the part being examined.

P1 (guided). A rights issue, end to end. Mahim Textiles Ltd (synthetic) has 25.00 crore shares at ₹288 and announces two new shares for every five held, at ₹180. Compute the new shares, the cash raised, the theoretical ex-rights price, the value of the right per new share and per existing share, and the loss to a holder of 1,500 shares who does nothing.

Solution. New shares = 25.00 × 2/5 = 10.000000 crore; cash raised = 10.00 × 180 = ₹1,800.000000 crore; shares after = 35.00 crore. TERP = (25.00 × 288 + 10.00 × 180) ÷ 35.00 = (7,200 + 1,800) ÷ 35 = ₹257.142857. Right per new share = 257.142857 − 180 = ₹77.142857. Right per existing share = 288 − 257.142857 = ₹30.857143, which also equals (288 − 180) × (2/5) ÷ (1 + 2/5). The lapsing holder loses 1,500 × 30.857143 = ₹46,285.714286.

P2 (guided). A recapitalisation and its stub. Cascade Fabricators Inc (synthetic) has 22.0 million shares at $41.00 and $90 million of net debt, on EBITDA of $96 million. It borrows $620 million and pays it out as a special dividend. Compute the dividend per share, the post-recap leverage, the stub, its effective leverage, and the stub price if enterprise value then falls 15%.

Solution. Enterprise value = 22.0 × 41.00 + 90 = $992.000000m. Special dividend = 620 ÷ 22.0 = $28.181818. Net debt after = $710m, or 710 ÷ 96 = 7.395833× EBITDA. Stub = (992 − 710) ÷ 22.0 = $12.818182, which is 282 ÷ 992 = 28.427419% of enterprise value and therefore 3.517730× levered. If enterprise value falls 15% to $843.2m, the stub is (843.2 − 710) ÷ 22.0 = $6.054545, a fall of 52.765957%. A 15% move in the business moved the equity 3.5 times as far, which is the multiplier working exactly as the arithmetic said it would.

P3 (guided). A note distributed in a deal. A merger pays part of the consideration in a seven-year senior note with a face of $100 paying 4.50% annually. The issuer's comparable seven-year debt yields 8.80%. Compute the note's value, its discount to face, and its value if the yield later falls to 7.00%.

Solution. Seven-year annuity factor at 8.80% = 5.066907; seven-year discount factor = 0.554112. Value = 4.50 × 5.066907 + 100 × 0.554112 = $78.212302, a discount to face of 21.787698%. At a 7.00% yield the note is worth $86.526776, a gain of 10.630648% on a 180 basis-point rally. Marking this note at face on the announcement date would overstate the consideration by $21.79 per note.

P4. Liquidation in miniature. Deccan Forge Ltd (synthetic) has 1.80 crore shares. Cash ₹40 crore at no haircut; receivables ₹180 crore at 25%; inventory ₹220 crore at 40%; land at a market value of ₹260 crore at 15%; plant at a net book value of ₹300 crore at 75%; goodwill ₹90 crore written off. Wind-down costs are 5% of realisations plus ₹45 crore. Liabilities total ₹415 crore. The shares trade at ₹41. Compute the liquidation value per share and the discount.

Solution. Realisations: 40.00 + 135.00 + 132.00 + 221.00 + 75.00 + 0.00 = ₹603.000000 crore. Wind-down = 0.05 × 603 + 45 = ₹75.150000 crore. Net = ₹527.850000 crore; less liabilities ₹415 crore leaves ₹112.850000 crore, or ₹62.694444 a share. At ₹41 the shares trade at a 34.603456% discount to the estate, an upside of 52.913279% if it were ever realised. The honest caveat is the plant haircut: at 60% rather than 75% the answer moves by more than ₹20 a share, so the number is a band.

P5. The one-dollar test. Over a decade a company earned $18.4bn, paid $5.2bn of dividends, repurchased $6.9bn of stock and issued $0.8bn. Its market capitalisation went from $21.0bn to $44.5bn. Compute the adjusted one-dollar ratio, the raw ratio, and the error from skipping the adjustment.

Solution. Retained = 18.4 − 5.2 = $13.200000bn. Adjusted value created = (44.5 + 6.9) − (21.0 + 0.8) = $29.600000bn, a ratio of 2.242424. The raw ratio is (44.5 − 21.0) ÷ 13.2 = 1.780303. Skipping the adjustment understates the ratio by 0.462121, because the buyback removed market capitalisation that was value returned to the holders who stayed. Note the direction: the raw calculation flatters a company that never buys back stock and penalises one that does.

P6. Promoter creep on a partial subscription. Nilgiri Ceramics Ltd (synthetic) has 40.00 crore shares and a promoter holding of 38%. It announces one new share for every four held. The public takes up 55% of its entitlement and the promoter takes its own entitlement plus the entire shortfall. Compute the promoter's stake afterwards and the creep in percentage points.

Solution. Promoter shares = 15.20 crore, entitlement = 3.800000 crore. Public shares = 24.80 crore, entitlement = 6.200000 crore, of which it takes 3.410000 crore, leaving a shortfall of 2.790000 crore. The promoter ends with 15.20 + 3.80 + 2.79 = 21.790000 crore of 50.000000 crore shares, a stake of 43.580000%, up 5.580000 percentage points. The public's decision not to fund, not the promoter's decision to fund, is what moved the register.

P7. Category and trigger. A US company trades at 34 times earnings, growing 17% with a dividend yield of 0.8%. Compute the ratio of price-earnings to growth and its dividend-adjusted variant, state the category, and compute the multiple implied by a sell trigger set at 2.0 if growth later decays to 9%.

Solution. The simple ratio is 34 ÷ 17 = 2.000000 and the dividend-adjusted variant is (17 + 0.8) ÷ 34 = 0.523529. Growth of 17% with a negligible yield puts it in fast-grower territory rather than stalwart. A trigger at 2.0 implies a multiple of 2.0 × 17 = 34.0× today, which the stock has already reached; if growth decays to 9% the trigger multiple falls to 18.0×, a 47.058824% price fall with no change in earnings at all. That second number is the one to write down, because de-rating and not disappointment is how most fast growers lose money.

P8. The four tests on a demerger. An Indian parent with 210 crore shares demerges a business at one new share for every two held. The new shares trade at ₹96 when-issued. Index-constrained holders own 29% of the parent. Expected daily traded value is 1.1% of market capitalisation. Compute the overhang in days of volume and say what score Test 1 earns.

Solution. New shares = 105.000000 crore; market capitalisation = 105 × 96 = ₹10,080.000000 crore. Forced holders own 0.29 × 10,080 = ₹2,923.200000 crore. Daily traded value = 0.011 × 10,080 = ₹110.880000 crore. Days of volume = 26.363636. That is a named, sized, mandate-driven seller holding about five and a half weeks of turnover, so Test 1 scores 2. It scores 2 on the size of the overhang and not on its direction. A large overhang tells you the selling is real and says nothing about whether the price it produces is below value, which is Test 2's job.

P9. Sizing a levered claim. A stub has four outcomes with probabilities 0.20, 0.30, 0.35 and 0.15 and returns of −100%, −35%, +55% and +160%. Compute the expected return, the full-Kelly and half-Kelly fractions, and the size a 2% ruin constraint permits. Say which binds and why.

Solution. Expected return = 0.20(−1.00) + 0.30(−0.35) + 0.35(0.55) + 0.15(1.60) = +12.750000%. Full Kelly is 18.839667% and half-Kelly 9.419833%. A ruin constraint capping the loss from any single position at 2% of capital, against a worst case of a total loss, permits 2.000000%. The ruin constraint binds, by a factor of 4.709917 against half-Kelly. Kelly is the growth-optimal answer for an investor who plays the same distribution an unlimited number of times and cannot be removed from the game after a bad run; neither condition describes anyone with clients, a career, or a finite life.

P10. Which anchor binds. Vindhya Alloys Ltd (synthetic) has 3.10 crore shares at ₹118, net debt of ₹480 crore, normalised EBITDA of ₹145 crore and normalised free cash flow of ₹58 crore growing 3.5%, and its cost of capital is 12.5%. Precedent transactions in the sector have been done at 7.2 times EBITDA. An asset-by-asset estate realises ₹355 crore net of wind-down and all liabilities. Compute the three anchors, say which binds, and compute the discount to it.

Solution. Private-market value = (145 × 7.2 − 480) ÷ 3.10 = ₹181.935484. Going concern = (58 × 1.035 ÷ 0.09 − 480) ÷ 3.10 = ₹60.322581. Liquidation = 355 ÷ 3.10 = ₹114.516129. The private-market anchor is highest and therefore binds any thesis that assumes a sale; the going-concern value is the lowest and is what you own if nothing happens; the liquidation value is the floor and it sits below the ₹118 price, so the market is not offering you the estate for free. The discount to the binding anchor is 35.141844%. The thesis has to name who buys it, because the ₹182 is only available through a transaction.

P11. Days of volume. A spin-off with a market capitalisation of $2.4bn is 36% held by index-constrained mandates. Its expected daily traded value is 0.75% of market capitalisation. Compute the overhang in days, and recompute it if actual volume comes in at twice the estimate.

Solution. Forced-seller value = 0.36 × 2,400 = $864.000000m; daily traded value = 0.0075 × 2,400 = $18.000000m; overhang = 48.000000 days. At twice the volume it is 24.000000 days. The ratio is one of the cleanest algebraic facts in event-driven work: the overhang in days is the forced share of the float divided by the daily turnover rate, so neither the market capitalisation nor the share price enters it. A note that quotes the overhang without the volume assumption has quoted half a number.

P12. The two-minute drill, with numbers attached. Barmer Auto Forgings Ltd (synthetic) has 4.20 crore shares at ₹690, revenue of ₹2,850 crore, an EBIT margin of 14.8% and a 25% tax rate. Compute earnings per share and the multiple, then compute the price at the same multiple if the margin reverts to 11.0%. Then write the four break conditions the drill demands.

Solution. Net income = 2,850 × 0.148 × 0.75 = ₹316.350000 crore; earnings per share = ₹75.321429; the multiple is 690 ÷ 75.321429 = 9.160740×. At an 11.0% margin earnings per share fall to ₹55.982143 and the price at the same multiple is ₹512.837838, a fall of 25.675676%. The four break conditions follow from that single sensitivity: the margin falling below 12% for two consecutive halves; the order book covering less than nine months of revenue; net debt to EBITDA rising above 2.0 times; and any customer concentration above 30% of revenue appearing in the segment note. All four are observable, all four are dated by the reporting calendar, and each one is a sell instruction written before the fact rather than a reaction after it.

Applied mini-project

Track one live corporate event for four weeks in the primary documents, and write the verdict before you know the answer.

Pick a corporate event that has been announced but not completed: a rights issue, a demerger or scheme of arrangement, a leveraged recapitalisation or large special dividend, a delisting or open offer, an insolvency resolution plan, or a deal paying part of its consideration in securities. The event must be live, and you must be able to reach its primary documents. Indian events are read on the exchange announcement pages, the company's own filings and the scheme or offer document; American events are read on the regulator's filing system.

Four artefacts, each with a test it has to pass before it counts.

The first artefact is the document trail, and it carries one test. Build a dated log of every primary document the event produced during your four weeks: the announcement, the offer or scheme document, the valuation report, any exchange observation letter, any clarification, and the results of any vote. For each, one line on what it added that the previous document did not. Test 1: every entry carries a document name and a date, and at least four distinct documents appear. A log built from news reports rather than filings fails.

The second artefact is the four-test scorecard, and it carries two tests. Score the event 0, 1 or 2 on each of the four tests, and show the arithmetic behind Tests 1 and 2. Test 2: the forced-seller overhang is expressed in days of the security's own average traded value, with the volume assumption stated. Test 3: the gap to value is computed against a named method, with the cost of carry, the expected holding period and the tax treatment subtracted, so the number you report is what you would actually keep.

The third artefact is the valuation, and it carries two tests. Value whatever the event distributes or leaves behind: the right, the stub, the distributed note or warrant, the spin-off, or the residual listed entity. Test 4: the valuation reconciles by an independent route, meaning the ex-rights price agrees with the value of the right, or the stub agrees with the pre-event price minus the distribution, or the sum of the parts agrees with the parent's market capitalisation plus the announced change. Test 5: every assumption is named with a number beside it, and the two that move the answer most are re-run at their pessimistic and optimistic ends, so the deliverable carries a band and not a point.

The fourth artefact is the price path and the verdict, and it carries one test. Keep a weekly price record for the security or securities involved, alongside the dated events from the first artefact. Then the verdict, written in two parts. Part one is written in week one, before the outcome: what you believe the mispricing is, in currency per share, and what will close it and by when. Part two is written in week four: what actually happened, whether the mispricing you identified was real, and what closed it or did not. Test 6: part one is timestamped before part two and is not edited afterwards. An unfalsifiable verdict scores zero for the artefact.

Deliverable. One workbook or notebook, plus a memo of no more than three pages carrying: the event and its terms in five lines; the four-test scorecard with the two computed tests shown; the valuation with its band and its two named sensitivities; the price path against the event dates; and the two-part verdict. Close the memo with one sentence on what you would do differently if you met the same event shape again.

The rubric runs to 20 points and passes at 16, with no zero on any row.

Rows are scored to 3 where the "3" standard is met and to 2 where only the "2" standard is met. Any artefact whose test fails scores zero for that row whatever the writing is like. The reason for that rule is the whole reason the project exists: a special-situations memo is a claim that a number is wrong, and a claim that does not reconcile is not evidence, however well it reads.

Reading & resources

The special-situations spine:

• Joel Greenblatt, You Can Be a Stock Market Genius (1997). The field guide to rights offerings, merger securities, recapitalisations, stubs and bankruptcies, written by someone who ran them. The chapters this node completes are the ones previously summarised in a sentence. [Paid] [Intermediate]
• Seth Klarman, Margin of Safety (1991). The three valuation anchors, the performance derby and the case for cash. Out of print and expensive second-hand; library copies and the widely circulated summaries are a legitimate substitute, and the argument survives compression better than most. [Paid] [Advanced]
• Peter Lynch, One Up on Wall Street (1989) and Beating the Street (1993). The six categories, the two-minute drill and the honest limits of the retail edge. [Paid] [Beginner]
• Philip Fisher, Common Stocks and Uncommon Profits (1958). The fifteen points and the scuttlebutt method are taught in the schools phase; the two chapters of don'ts are what section 4.14 rebuilds. [Paid] [Intermediate]
• Warren Buffett, Berkshire Hathaway shareholder letters, particularly 1984 for the retained-earnings test and the owner's manual for the standard restated. Free on the company's own site, and the primary source for the one-dollar test. [Free] [Beginner to Advanced]

Primary documents to practise on:

• SEBI and the exchanges (NSE, BSE) corporate-announcement pages, for letters of offer, scheme documents, valuation reports, exchange observation letters and voting results. An Indian event lives here and nowhere else. [Free] [Beginner]
• SEC EDGAR full-text search, for Form 10 registration statements on spin-offs, merger proxies, warrant agreements and indentures. A Form 10 is the single best free document for practising Test 4. [Free] [Intermediate]
• The Insolvency and Bankruptcy Board of India and the tribunal's own order pages, for admitted cases, resolution plans and approved outcomes; the place to check whether listed equity survived a resolution rather than assuming it did. [Free] [Intermediate]
• Index provider methodology documents (NSE Indices, S&P Dow Jones, FTSE Russell, MSCI), which state exactly what a tracker must do with a spin-off and by when. Test 1 is a reading exercise in these before it is an arithmetic exercise. [Free] [Intermediate]

Adjacent modules that own what this one cites:

• E11.02 M&A, Deals & Special Situations Lab for spin-off and demerger mechanics, sources and uses, and merger-arbitrage spread math.
• E11.01 Credit Analysis & the Debt Investor's Lens for the recovery waterfall, covenants and the fulcrum security from the credit side.
• E11.04 Governance & Forensics: India Lab II for promoter mapping, related-party materiality and the pledge and auditor signals behind the governance gate.
• E11.05 Market-Level Valuation, the Index Default and Your Own Policy for the allocation band, the rebalancing rule and the index default that decide how much capital is available for anything here.
• M3.10 Special Situations + Phase Capstone for sum-of-the-parts, the holding-company discount, distress and equity as an option.
• M6.01 The Schools I and M6.02 The Schools II for Graham, Buffett, Lynch, Greenblatt, Klarman and Fisher as schools rather than as drills.

Do this, not just read: take one Form 10 or one Indian scheme document and, without looking at any price, write the four-test scorecard from the document alone. Then look at the price. Most of the value in event-driven work is created in the gap between those two acts, and doing them in the wrong order is how a thesis becomes a rationalisation.

The Modern Analyst's Addendum

The teaching above is done by hand, and the mastery check still tests it that way. What follows is how a working analyst amplifies the same skill without outsourcing the judgement.

AI-Augment this skill

{
  "skill": "Extracting the terms of a live corporate event from its primary documents, and building the four-test scorecard and the valuation the event needs",
  "use": "The honest use here is clerical and it is genuinely large, because event documents are long, structured and public. Have an assistant pull the terms out of a letter of offer, a Form 10, a scheme of arrangement or a warrant agreement into a table you then check clause by clause: the ratio, the subscription or exercise price, the record and entitlement dates, the renunciation window, the promoter's stated intention on the unsubscribed portion, the anti-dilution and adjustment clauses, and the approvals required with their thresholds. Have it turn an index provider's methodology into a plain checklist of what a tracker must do with the new security and by when, which is the input Test 1 needs. And have it draft the skeleton of the valuation you will populate: the ex-rights reconciliation, the stub ladder, the liquidation schedule with a haircut column left blank. Never ask it for the haircuts, the probabilities, the peer multiple or the score on Test 4, because those are the judgements the whole exercise consists of.",
  "tools": ["Chat assistants with the filing text pasted or uploaded, for term extraction from a letter of offer, Form 10, scheme document or warrant agreement", "SEC EDGAR full-text search and the NSE/BSE corporate-announcement pages, for the source documents themselves", "Python with a spreadsheet or notebook, for the ex-rights reconciliation, the stub ladder, the option value and the liquidation schedule", "An index provider's own methodology PDF, as the primary source behind any forced-selling estimate"],
  "prompt": "From this [letter of offer / Form 10 / scheme of arrangement / warrant agreement] excerpt, extract into a table, one row per field: the entitlement or exercise ratio; the subscription or exercise price; the record date, the entitlement credit date and the closing date; whether rights entitlements are renounceable and over what window; any stated intention by the promoter or a controlling shareholder regarding its own entitlement and regarding the unsubscribed portion; the stated objects of the issue, item by item with amounts; every approval required, with its voting threshold and whether interested parties are excluded; and every anti-dilution or adjustment clause. Quote the exact clause reference for each field. Write NOT STATED rather than inferring anything, and list separately every figure in the document you could not locate.",
  "verify": "Open the clause the tool quoted for the promoter's intention on the unsubscribed portion and for the objects of the issue, and read both yourself, because those two fields decide the governance question and the investment question respectively and an extraction error in either changes the answer completely. Recompute the ex-rights price and the value of the right by both routes and require them to agree to the paisa; if they do not, a share count came out of the document wrong. Trace the entitlement ratio and the record date to the exchange announcement rather than to a news report. Date-check every document against the event calendar, because a superseded offer document extracts perfectly and is worthless. And treat any number the tool supplies that is not in the document, a haircut, a peer multiple, an implied probability, as absent rather than as an estimate: the model has no source for it and will produce a plausible one on request.",
  "diy": "The gate is unaided. You compute a theoretical ex-rights price and the value of a right by two routes and reconcile them; you work the promoter's stake under full subscription, partial subscription with the shortfall absorbed, and a scaled-down issue; you price a distributed note off a comparable yield and a distributed warrant with the dilution adjustment; you build a stub ladder and size it against a ruin constraint; you run a liquidation schedule with your own haircuts and produce a band; and you compute a one-dollar ratio with the buyback adjustment. No extraction tool is in that room, and none of that work needs one."
}

Modern Data Analysis

By hand first. Every worked example above is a single snapshot: one price, one set of haircuts, one peer multiple, one probability vector. That is the right way to learn it and it is frozen the moment the spreadsheet closes, because the when-issued price reprints tomorrow and the haircut was always a judgement.

Today's workflow. Turn each of the four computations into a function of its inputs, then sweep the input that is an assumption rather than an observation. The liquidation schedule becomes a function of a haircut vector, and the useful output is the band across a plausible range rather than the point estimate. The stub becomes a function of the enterprise-value distribution, and the useful output is the position size implied by a ruin constraint at each candidate price. The overhang becomes a function of the volume assumption, and the useful output is the number of days at the tenth and ninetieth percentile of the security's own recent turnover rather than at a round number somebody picked. And the one-dollar test becomes a function of the window, which is how you discover, as the worked example did, that a year of endpoint moved the verdict by a third.

Tools and sources. numpy and pandas for the sweeps, scipy.optimize for the implied-volatility and implied-yield inversions if you would rather not bisect by hand. For India: exchange announcement pages for the documents, and daily traded-value history for the volume assumption behind any overhang estimate. For the United States: the regulator's full-text search for the Form 10 and the warrant agreement, and the index provider's methodology for the reconstitution dates that Test 3 depends on. No vendor feed is required for any of the arithmetic; the inputs are all public and the work is in checking them.

Quantitative lens

Three of the numbers above are far more sensitive to their assumptions than they look, and knowing which is the difference between a band and a bluff.

The liquidation value is the worst offender. The central estimate of ₹126.15 becomes ₹48.34 at pessimistic land and plant haircuts and ₹191.69 at optimistic ones, a band of ₹143.34 that is wider than the estimate itself. Two of the thirteen lines account for almost all of that, and they are the two that are hardest to observe. Anyone publishing a liquidation value without those two sensitivities has published a number whose error bar is larger than its value.

The overhang is the best behaved, and the algebra says why. Days of volume equals the forced share of the float divided by the daily turnover rate, so the share price and the market capitalisation cancel out entirely. Only two inputs matter, one of which is published in the index methodology and the other of which is estimated from the security's own recent trading. Doubling the volume assumption exactly halves the answer, which makes the sensitivity trivial to state and impossible to hide.

The one-dollar ratio sits in between and its sensitivity is structural rather than statistical. The numerator contains two market capitalisations, and the market's own error at either endpoint flows straight into the answer. Ending the synthetic decade one year early moved the ratio from 2.82 to 1.86. That is why the test needs a decade, why it is most useful when the answer is far from 1.0, and why an analyst who reports it should report the window alongside it as a matter of course.

Verify block: every printed figure, at full precision

Teach it back & journal

Feynman prompt. Write a one-page explanation, for a friend who owns two mutual funds and has never read a filing, titled "The letter you did not open cost you ₹52,000." Use no formulas and no jargon. You have to get three things across. First, that when a company offers its existing owners new shares cheaply, the cheapness is not a gift, because the share price adjusts the moment the new shares exist; the person who acts keeps what they had, and the person who ignores the letter hands it to somebody else. Second, that the person most able to act is usually the family that controls the company, which is why an offer nobody takes up quietly moves ownership. Third, that the only question worth arguing about is what the company will do with the money, because that is the part the arithmetic cannot fix. Then, in two sentences at the end, say what you would actually tell your friend to do the next time she gets one of those letters. If you cannot write those two sentences without hedging, you have not finished the first three points.

Journal prompt. Open your decision journal and write an entry called the four tests, applied to something I already own. Take one position in your portfolio, or one from a watchlist if you hold nothing yet, and score it on the four tests as though it were an event-driven situation. Can you name a seller, and can you size the gap net of what it costs to wait? Then ask whether a dated thing closes it, and what the person selling to you knows that you do not. Most ordinary positions score badly on all four, and that is the useful finding, because it tells you that the position is an opinion about a business rather than a claim about a mispricing, and those two things deserve different sizes and different holding periods. Then write the four break conditions from the two-minute drill for that same position, as observable numbers with reporting dates attached, and date the entry. Two things to check when you reread it: whether any of your four conditions was unfalsifiable, and whether you sized the position as an opinion while describing it as an edge. Come back to the entry after the next set of results and mark, honestly, which of the four conditions you actually checked.