Investing Classics: Shorting & Special SituationsReturns That Come From Structure, Not Growth

The misattribution is itself the warning: widely repeated is not the same as verified. Shorting carries three asymmetries. The odds are inverted (capped at 100%, uncapped below). Position feedback is positive: being right shrinks the position and makes compounding hard, while being wrong inflates it and reinforces the risk. Holding costs money: borrow fees, dividends paid away, recall of the borrow at any moment — and equities drift up over time. "Expensive" is never a reason to short; you need a catalyst and a timetable.

In November 2000, Jim Chanos's Kynikos began shorting Enron on evidence drawn entirely from public filings: a return on capital of only 6%–7%, below its cost of capital, plus gain-on-sale revenue recognition and extensive related-party transactions. Enron filed for bankruptcy on December 2, 2001; in February 2002 Chanos laid out the same reasoning in testimony to the U.S. House of Representatives.

The reverse: on October 26, 2008, Porsche disclosed holdings of 42.6% of Volkswagen plus 31.5% in cash-settled options — roughly 74% — with Lower Saxony holding another 20.1%. That left under 6% genuinely tradable, against short interest of about 12%. Two days later the shares ran from roughly €210 to above €1,000. The shorts' valuation judgment was not wrong; the other side was simply able to squeeze them out first.

In the January 2021 squeeze, Melvin Capital lost about 53% in a single month and wound down the following year despite a $2.75 billion injection. The deeper point: even Chanos closed his outside fund business in November 2023, on the grounds that standalone short selling had become unsustainable as a business. Shorting is excellent research training and rarely a good position.

• Does my thesis have a catalyst and a time window, or only "it's too expensive"?
• What share of the float is short? What is the annualized borrow rate? Can the borrow be recalled?
• If the price doubles first, can I avoid a forced close and live to see the thesis play out?

Everything turns on the fourth question: the first three determine your return, the fourth determines whether you survive. The spread is not a discount; it is the market's quoted price for the probability of failure. An annualized 4% spread implies that better than nine-in-ten completion odds are already priced, and the remaining tenth takes 20%–30% in one stroke. In the same letter Buffett notes that Graham-Newman's arbitrage returned roughly 20% a year over three decades — in an era when arbitrageurs were scarce.

Microsoft's acquisition of Activision Blizzard was announced in January 2022 at $95 per share in cash. The FTC sued that December and the UK's CMA blocked it in April 2023; the shares fell to about $71, a discount of roughly 25%. The deal closed in October 2023 — still at $95. One transaction, opposite outcomes for holders with different horizons and different leverage.

The other side: in April 2016 the U.S. Treasury abruptly issued new rules on corporate inversions, and the roughly $160 billion Pfizer–Allergan merger was terminated within two days. The rule change came without warning and sat in no model.

The antitrust environment is an institutional variable, not a probability constant: the FTC/DOJ and the UK's CMA visibly tightened their posture in 2021–2023, and the prior decade's break rates were worthless as guidance for that stretch. An individual also carries three disadvantages: no institutional funding cost, no ability to spread across dozens of concurrent deals, and no sight of what law firms judge about the approval path — and those three are the moat of the entire business.

• Am I measuring the annualized spread or the absolute one? What remains after tax and funding costs?
• What is the "naked" price on a break (pre-announcement price × sector move since)? How many such losses can I absorb?
• Is the main risk the shareholder vote, financing, or antitrust? The last cannot be estimated from historical odds.

The absolute priority rule sets the waterfall: secured debt → unsecured debt → preferred stock → common stock, with each layer entitled to nothing in principle until the one above is paid in full. The fulcrum security is the layer the line happens to cut through, and after reorganization it typically converts into equity in the new company. So the work has two steps: estimate the reorganized value of the enterprise, then see which rung of the waterfall that number lands on. It also explains why old common stock so often spikes on a bankruptcy filing — most of the buyers do not know they are standing at the bottom.

General Motors filed for Chapter 11 in June 2009. Secured lenders were paid in full; holders of roughly $27 billion of unsecured bonds received equity and warrants in the new company worth about thirty cents on the dollar initially; the old common stock went to zero. It is also a case of absolute priority being broken — the UAW retiree healthcare trust (VEBA) was treated better than some pari passu unsecured creditors. Priority is the legal starting point; the outcome is settled by negotiation.

The other side: in the week Lehman failed, Howard Marks of Oaktree began buying distressed debt in size, deploying roughly $6 billion over the following fifteen weeks. His reasoning was not that he had called the bottom, but that prices already embedded a collapse of the system — and if that actually happened, nothing anyone did would matter.

Unsecured creditors of Lehman Brothers Holdings ultimately recovered far more than early estimates suggested — but the liquidation ran for more than a decade: the recovery rate was right and the annualized return still poor. The more practical limit is access rather than difficulty: corporate bonds often trade in $100,000 minimums, court filings and creditors' committees create structural information gaps, and legal and advisory fees only make sense at scale.

• How many billions of senior claims sit above the layer I am buying?
• Where is the fulcrum security? Am I above it (cash or new debt) or below it (likely zero)?
• If the reorganization takes three years, what is left of my annualized return?

There are three observable sources of excess return. Forced selling: spin-off shares are distributed pro rata to parent shareholders, many of whom (broad index funds, large-cap-only institutions) cannot or will not hold them. An information vacuum: no analyst coverage and no standalone financial history. Reset incentives: management receives equity tied directly to its own business for the first time. Greenblatt cites a Penn State study finding that over the twenty-five years to 1988, spun-off companies beat the market by roughly 10 percentage points a year in their first three years.

The 1993 Marriott split is the canonical example: the asset-light, cash-generative hotel management business went into Marriott International, while the real estate and the bulk of the debt stayed with Host Marriott. Existing bondholders sued — their claims had been left on the weaker side of the assets. The dispute is precisely the lesson: the first purpose of a spin-off is often to redistribute debt and risk, not to release value. Whoever takes the debt is the side that may be mispriced.

Today the mechanism is fully anticipated: GE split into three companies across 2023–2024 and Johnson & Johnson spun off Kenvue in 2023, both studied broadly months in advance. Mechanically buying spin-offs is no longer a free lunch.

This is the textbook case of anomaly decay. McLean and Pontiff (2016, Journal of Finance) examined 97 published return predictors and found that returns fell by roughly 58% on average after publication; an effect written into a bestseller decays faster still. The other face is just as common: parents use spin-offs to hand off declining businesses, pension liabilities, or litigation — you think you have bought a neglected pearl and have in fact taken delivery of what was being discarded.

• Which side received the debt, pension, and litigation liabilities? What is the spun-off entity's net debt/EBITDA?
• Which side did the core management team go to? What metric does their new equity award track?
• Is my reason "structural selling pressure" or "this is a good business"? The second requires an entirely different analysis.