Short Selling
Short Selling: Strategies, Risks, and Rewards
Eighteen contributors, one editor, no single argument. The mechanics chapter explains borrow, rebate rates and recall risk better than most trading books bother to, and Owen Lamont's chapter on 3Com and Palm is the best short case study in print. The other eleven chapters are portfolio theory.
The one idea
A stock stays overpriced when the people who know it is overpriced cannot bet against it. That is the whole book in one line, and it comes from a 1977 paper by Edward Miller that this collection builds on for 400 pages. Buying is easy. Anyone with a brokerage account can push a price up. Selling short requires you to first find someone who owns the stock and will lend it to you, post collateral, keep posting more collateral as the price moves against you, and hand the shares back whenever the lender asks. Every one of those steps is a place where the trade dies. When enough of them fail at once, the pessimists are locked out and the price is set entirely by the optimists.
The book's best evidence is a natural experiment from March 2000. 3Com sold a slice of its Palm subsidiary to the public and announced it would spin off the rest, with each 3Com share entitled to about 1.525 Palm shares. That means 3Com had to be worth at least 1.525 times Palm, since a share of stock cannot be worth less than zero. Palm closed its first day at $95.06, so 3Com should have traded above $145. It traded at $81.81. The market was pricing 3Com's profitable networking business, plus more than $10 a share in cash, at negative $22 billion. The trade to fix it was obvious and was covered in the Wall Street Journal, the New York Times, and USA Today the next morning. Almost nobody could put it on, because Palm was nearly impossible to borrow.
That is what the book is for. It is not a manual for making money short. It is a manual for understanding why the money is still sitting there. The practical payoff is defensive: once you know what the borrow market actually costs and how it fails, you stop treating a short as a long trade with a minus sign in front of it.
What it actually teaches
The book never lays out a procedure, so here is the sequence its chapters describe, in the order a trade actually happens.
DECIDE IF YOU NEED THE BORROW AT ALL
Fabozzi's chapter three and Gastineau's chapter four both argue you often do not. Index futures let you short a whole market with no locate and no recall. Buying puts or writing calls does the same for a single name. Exchange-traded funds are the cleanest of the three: a squeeze is effectively impossible because an Authorized Participant can create new ETF shares on any trading day, which is why short interest in large ETFs runs 20% to 55% of shares outstanding while the average listed stock sits near 2%. Gastineau also notes ETFs are exempt from the uptick rule and carry no roll risk, unlike futures, which get rolled about four times a year. Caution on chapter three: it devotes several pages and a full contract-specification exhibit to single-stock futures on NQLX. Those products are gone.
LOCATE THE BORROW
Cohen, Haushalter and Reed open by admitting the equity lending market is arcane and that most loans are negotiated over the phone with no electronic quote network. You must find a lender before you sell. Custodian banks lending on behalf of pension funds are the traditional supply, typically splitting revenue 75% to the beneficial owner and 25% to the agent bank. Hedge funds cannot borrow directly because lenders will not approve their credit, so they go through a prime broker. The chapter states plainly that for stocks in genuine demand, loans are expensive for well-placed investors and impossible for retail. Read that sentence twice before building a plan around it.
POST THE MARGIN AND THE COLLATERAL
Two separate money requirements, and readers confuse them constantly. At your broker, Regulation T requires 50% initial margin, and the self-regulatory organizations require you to maintain at least 30% of the market value of the short position as the price moves. Separately, at the lender, the sale proceeds are handed over as collateral at 102% of the loan value for US stocks, marked to market daily, so a rising price triggers a call for more collateral. Jacobs and Levy note lenders in practice usually demand around 105%. They also recommend holding roughly 10% of your capital in cash as a liquidity buffer purely to meet those daily marks and to pay the dividends you now owe the lender.
PRICE THE TRADE OFF THE REBATE, NOT A COMMISSION
You do not pay a borrow fee in the normal sense. The lender invests your collateral and rebates part of the interest back to you, and your real cost is the gap between the market rate and that rebate. For liquid large caps you get the general collateral rate, roughly 5 to 25 basis points below Fed funds, widening to about 35 basis points for mid caps. Roughly 7% of stocks trade 'special' on any given day, and 2.77% of large caps do, meaning the rebate is negotiated case by case and can go deeply negative. The book's examples: a new IPO averages 300 basis points below general collateral, falling to 150 within 25 trading days; merger acquirers average 23 below; Stratos Lightwave hit 4,000 basis points below general collateral just after its August 2000 IPO. Lamont's chapter puts Palm at a 35% annual borrow cost in July 2000, while Palm's option prices implied a holding cost of 119% annualized over two months and 147% over eight.
FIND THE TARGET IN THE FILINGS
Chapter ten is the only chapter written by people running a live short book, Ron Gutfleish and Lee Atzil of Elm Ridge. Their thesis: buy stress and short comfort, because analysts covering a beloved story stop being vigilant. Their eight screens, run on seasonally adjusted quarterly numbers, are a growing gap between net income and cash flow from operations, receivables growing faster than revenue (rising DSO), inventories growing faster than sales (rising DSI), other current assets rising against revenue, depreciation falling relative to gross property plant and equipment, declining allowances for doubtful accounts and returns and inventory obsolescence, keyword hits in filings for 'reversals', 'off-balance sheet', 'special purpose entity', 'related party', and 'previously' or 'change' near 'recognized', and finally a falling ratio of shares short to float. That last one is a squeeze filter, not a fundamental one. Screens only tell you where to look. They then insist you identify the bull case and what breaks it, because a red flag nobody cares about pays nothing.
SIZE FOR THE PAIN, NOT FOR THE THESIS
Elm Ridge runs up to 120 short positions at once, sized inversely to volatility. A volatile tech short where they expect 50% gets a smaller weight than a consumer staples name with limited upside in both directions. They deliberately keep dry powder because they expect to be early and to take at least one more bad quarter, and they scale in as the position moves against them. Their words: they need to make sure positions do not cause so much pain as to provoke extreme irrationality and panic. Note the honesty and note the danger: adding to a losing short is exactly the behavior that ends accounts, and their own third example shows the stock tripling on them.
MANAGE RECALL AND SQUEEZE UNTIL YOU COVER
Recall happens in roughly 2% of loans. When it does, the agent sends notice on T+1 and you have until T+3 to return shares, or the lender buys in against your collateral at whatever price the market offers. You have no say in who gets recalled; the agent picks, then the prime broker picks again. Squeezes come in two forms. The original one is mechanical: holders pull shares out of the lending pool so borrowers must cover. Lamont's Solv-Ex case is the textbook example, where management faxed shareholders on 2/5/96 urging them to take delivery of certificates, and the stock went from $24.875 on 2/2/96 to $35.375 by 2/21/96 before the company was later found to have defrauded investors. The modern one is just price: shorts panic and cover, which lifts the price, which panics more shorts. Jones and Larsen close their chapter with a blunt line worth taping to a monitor: the only reason to buy or hold a stock with high short interest is if you think a squeeze is coming.
What it looks like Monday morning
You stop quoting short interest as a signal. Jones and Larsen spend a full chapter on 25 years of academic work and conclude the evidence is mixed and the signal is weak. Large percentage increases in short interest weakly predict negative short-term returns; accumulated relative short interest is a stronger long-horizon signal, but the relationship weakened after 1994 as hedge fund hedging and arbitrage muddied the data. Options trading in a name obscures it further. Meanwhile Lamont shows short interest can move opposite to overpricing: Palm's short interest peaked at 147.6% of all shares issued on 7/14/00, at the exact point the mispricing was already correcting.
The other change is procedural. Before you build a short thesis, you call your broker and find out whether the stock is available and what the rebate is, because the answer decides whether the idea is tradeable at all. If it is on special, you are competing with people paying tens of percent per year to hold the same view, and the book's own authors say retail orders in special names simply get denied. If it is not borrowable, you check whether puts or a sector ETF express the same view. And you calculate your maintenance margin at a price 50% above your entry before you put the trade on, not after.
| Component | Market rises 30% | Market falls 15% |
|---|---|---|
| $9M long positions | up 33%, gain $2.97M | down 12%, loss $1.08M |
| $9M short positions | up 27%, loss $2.43M | down 18%, gain $1.62M |
| Net equity result | gain $540,000 | gain $540,000 |
| Interest: 5% short rebate on $9M | $450,000 | $450,000 |
| Interest: 5% on $1M cash buffer | $50,000 | $50,000 |
| Ending value on $10M | $11.04M, up 10.4% | $11.04M, up 10.4% |
The same 10.4% in a bull and a bear market, but 5.0 of those 10.4 points is just the interest rate, which is the half that disappeared once short-term rates went to zero.
Stocks are only overpriced when informed investors are unable or unwilling to short them.Owen A. Lamont, Short Selling: Strategies, Risks, and Rewards
Where it fails
- The risk chapters end four years before the risk arrived. Lamont's history of political attacks on short sellers is genuinely good: the NYSE's special rules in November 1917 over fears the Kaiser was driving prices down, the 1929 crackdown that produced the uptick rule and the Investment Company Act of 1940, Malaysia's Finance Ministry proposing caning for short sellers in 1995. He treats it as a recurring nuisance. Then in September 2008 the SEC banned short selling outright in 799 financial stocks and the UK regulator did the same, which is not a nuisance, it is your entire book being closed by decree while you hold it. Nothing in this 2004 edition prices that. There is no chapter, no paragraph, no worked example where the regulator turns the trade off.
- The squeeze math is calibrated to a world that no longer exists. The book's two headline squeezes are Solv-Ex, up about 42% over nineteen days in 1996, and Martha Stewart Living, from just over $9 to $13.39 between mid-December 2003 and the end of January 2004 on more than 50% of float short. Those are the extremes it teaches you to survive. In January 2021 GameStop went from the high teens to $483 intraday in about three weeks and took a multibillion-dollar fund with it. Read Gutfleish and Atzil's advice through that lens: they tell you to scale into a losing short when the market is telling you that you are wrong. In 2003 that was disciplined. In a coordinated retail squeeze it is a liquidation. The book has no concept of a squeeze driven by an outside crowd rather than by lender recall, because that mechanism had not yet appeared at scale.
- Several pages of chapter three describe a product that was delisted. Fabozzi reproduces the full NQLX single-stock futures contract specification, including settlement price rules and position limits, and names three specific advantages over borrowing stock: no locate problem, no recall risk, and often a lower net interest cost. NQLX shut down US operations shortly after publication and OneChicago, the last US single-stock futures venue, closed in September 2020. The advantages were real. The instrument is gone. A reader who follows this chapter is shopping for something that is not on the shelf.
- Roughly 40% of the page count is one professor's theory, and none of it is tradeable. Edward Miller wrote chapters 5, 6, and 14, about 167 of the roughly 393 pages of text. The editor thanks him for it in the preface. The material is the intellectual foundation of the whole collection and it is genuinely important, but it is divergence-of-opinion theory about why beta is underrewarded and why closed-end funds trade at discounts. It contains no instructions. If you bought this book for the subtitle's promise of strategies, you paid for 167 pages of academic finance to get to Gutfleish and Atzil's twenty.
- The one working practitioner chapter withholds every company name. Gutfleish and Atzil explain why, and the reason is legitimate: management teams of shorted companies initiate squeezes and lawsuits. But it means you cannot verify a single claim. You get an anonymous IT outsourcer whose EPS came in at one fifth of guidance, an anonymous book publisher covered 50% lower, and an anonymous consumer electronics component maker that tripled against them. Credit where it is due, they lead you through the loser in as much detail as the winners and admit the mistake was waiting one more quarter after a clean report. That is more honest than most trading books manage. It is still three unfalsifiable anecdotes.
- It quietly assumes you are an institution, and never says so up front. The mechanics chapter states that loans in special stocks are expensive for well-placed investors and impossible for retail, and that brokers will simply deny the short order. Jones and Larsen add that rebate rates are usually not available to individual investors at all. Lamont notes brokers can only lend from margin accounts, not cash accounts. So the exact stocks the book's central theory identifies as most overpriced are the ones a retail reader is structurally barred from shorting. That is a serious limitation on who this book can help, and it is buried in the middle of chapters rather than stated in the preface.
- Half of the market-neutral return in the worked example was the interest rate. Jacobs and Levy's $10 million account earns 10.4% in both a bull and a bear market, and 5.0 of those 10.4 points is a 5% short rebate plus interest on the cash buffer. That was reasonable in 2004. Through most of 2009 to 2021 short-term rates sat near zero and that half of the return simply vanished, leaving only the long-short stock selection spread. Gastineau's own footnote foreshadows it: in summer 2003 market makers pulled back from lending ETF shares because rates were too low to make it worth doing. The structure still works. The advertised number does not travel.
- Small mechanics have drifted since this printing. This edition is written throughout in T+3 settlement terms, including the recall timeline where the agent notifies on T+1 and demands shares back by T+3. US equities moved to T+2 in 2017 and T+1 in May 2024, which compresses that window. The uptick rule the book discusses at length, and which Gastineau calls an anachronism, was repealed in 2007 and replaced in 2010 by Rule 201, a circuit-breaker that only restricts shorting after a stock falls 10% in a day. Regulation SHO and its locate requirement came in 2005, after publication. The 50% Reg T initial and 30% maintenance figures still hold. Check every rule citation in this book against a current source before you rely on it.
Who it is for
Buy it if
You already short stocks, or run a long-short book, and you have been treating the borrow as an afterthought your prime broker handles. Chapter two is the clearest short explanation of rebate rates and recall risk anywhere, and chapter seven will change how you think about what a crowded short actually signals. Buy it used, read three chapters, keep it as a reference.
Skip it if
You want a screen you can run tomorrow that produces short candidates. Chapter ten is the only chapter that comes close, it is twenty pages long, and its methods are widely available elsewhere in more detail. Skip it too if you are a retail trader with a small margin account, because the book's own authors state that the stocks its theory identifies as most overpriced are ones your broker will not let you borrow.
As an Amazon Associate I earn from qualifying purchases. It costs you nothing extra, and it does not change what gets recommended or what the verdict says.