1
Monday, October 19, 1987. 7:00 a.m. The New York Stock Exchange (NYSE).
18 Broad Street, at the southern tip of Manhattan. Six Corinthian columns at the Exchange's main entrance stood against a sky still tinged with darkness. Completed in 1903, the neoclassical facade — designed by George B. Post — was the face of Wall Street. Cold morning air cut between the columns. October dawns in New York are sharp. The smell of fog rolling up from the Hudson mingled with the scents of a Wall Street morning — steam rising from coffee carts, taxi exhaust, the ink of newspapers discarded on the pavement. That morning, the front page of The New York Times led with Friday's 108-point plunge.
On an ordinary day, the Exchange floor two hours before the opening bell is quiet. Across the expansive hall lined with seventeen horseshoe-shaped trading posts, specialists sip coffee and organize their ledgers. The floor spans roughly 3,300 square meters — about half the size of a football pitch. The ceiling soars twenty-two meters high. When light pours through the arched windows, it falls across the wood and metal surfaces of the trading posts in a muted glow. In this cathedral-like space, the daily rites of capitalism begin at 9:30 a.m. Each post is octagonal or horseshoe-shaped: specialists, clerks, and phone operators stationed inside, while floor brokers shout orders from the outside. Ticker symbols and last-trade prices blink on the electronic boards above each post.
This morning was different.
There were too many people on the floor. The opening bell was still two and a half hours away, yet traders were arriving in droves. Their faces were drawn. Everyone knew what had happened over the weekend.
On Friday, the Dow Jones Industrial Average had fallen 108 points in a single day — a 4.6 percent decline and the first 100-point drop in history. From Wednesday through Friday of the previous week, the cumulative loss had reached 250 points. In one week, the Dow had shed 17 percent from its all-time high. And over the weekend, markets across Asia and Europe had collapsed. The chronology unfolded like this: after New York's Friday plunge, the first market to open on Monday morning — crossing the International Date Line — was Sydney. Australian equities fell 3.7 percent; the bleeding had begun. Tokyo opened next; the Nikkei dropped 2.5 percent. Then came Hong Kong. The Hang Seng Index crashed 11.1 percent in a single day, and exchange chairman Ronald Li declared a four-day market closure "to protect investors." A closed market does not contain fear — it ferments it. When the Hong Kong Exchange reopened a week later, the Hang Seng plunged an additional 33.3 percent. London's FTSE 100 fell 10.8 percent. As the Earth completed one rotation, panic was migrating westward along the time zones: from Sydney to Tokyo, Tokyo to Hong Kong, Hong Kong to London.
Now it was New York's turn.
That morning, the NYSE opened its DOT (Designated Order Turnaround) system an hour early, giving brokerage firms extra time to enter their orders. In that single hour, firms loaded the system with four to five times the normal volume. The vast majority were sell orders. Before the opening bell rang, sell orders stacked in the DOT system amounted to 14 million shares, roughly $500 million in value.
The specialists' "books" — ledgers recording buy and sell orders — were crammed with sell orders that had accumulated over the weekend. Buy orders were virtually nonexistent. A specialist is the designated market maker for a given stock, responsible for facilitating trades and providing liquidity. When the market tilts overwhelmingly in one direction, the specialist is obligated to stand on the opposite side and restore balance — to buy when everyone else is selling. NYSE rules imposed on specialists the duty to maintain an "orderly market." If there were no buyers, they had to purchase shares with their own capital. This morning, that obligation weighed on them with unprecedented force. They were being asked to hold back a waterfall with their own money.
John Phelan, chairman of the NYSE, received a phone call before he had even left for the office. A former Marine, Phelan had served as NYSE president since 1980 and as chairman and CEO since 1984. A trader by background, he had an instinct for reading the mood of the floor. The same question came from the SEC and the White House: "Should we open the market?" After the collapse across Asia and Europe, there was genuine concern that opening New York could trigger a global chain reaction. Hong Kong had already shut its market. Shouldn't New York do the same?
Phelan's answer was unequivocal: "If we close the market, we may never be able to reopen it."
He understood market psychology. If the Exchange closed, sellers would be unable to sell, fear would find no release, and when trading resumed, a sell-off far worse than the current one would crash down on the market. Hong Kong would prove him right. The market had to open. Only an open market can discover prices, and only discovered prices can find an equilibrium. That is the logic of markets. In theory, at least.
Before the opening bell, Phelan summoned the CEOs of the major brokerage firms to his office. He assessed the volume of sell orders that had accumulated since Friday's crash and reviewed each firm's readiness. He connected by phone with White House Chief of Staff Howard Baker and reached out to the newly installed Federal Reserve Chairman Alan Greenspan. He called Gerald Corrigan, president of the Federal Reserve Bank of New York, to confirm the state of credit supply from the banks. He was preparing for the aftermath before the market had even broken. "Under no circumstances will the Exchange close. If extreme order imbalances arise in individual stocks, we will halt trading in those stocks, display new quotes, and resume. But the Exchange itself will not close." That was his directive.
While they waited for the opening bell, an abnormal hum pervaded the floor. Not shouting — murmuring. Hundreds of traders speaking in low voices all at once. The sound of printers connected to the DOT system spitting out paper — zzzzzt, zzzzzt, without pause. Phones ringing ceaselessly. It was the drone of a hive moments before it is disturbed. The traders' multicolored mesh jackets — yellow, red, blue, green, each color denoting a different firm — churned together through the crowd. Normally, this would be the hour for jokes and banter. This morning, no one was laughing.
9:30 a.m. The opening bell rang.
2
The moment the market opened, the ticker began falling behind.
Eleven of the thirty Dow Jones stocks failed to begin trading. Ninety-five stocks in the S&P 500 were unable to open. Sell orders were stacked up, but there were no buy orders, leaving specialists unable to set opening prices. A trade requires both a buyer and a seller. A market with only sellers is not a market at all. NYSE rules granted specialists the discretion to delay an opening, with a floor official's permission, when order imbalances were severe. But this morning, the delays were not isolated. Blue-chip stocks, one after another, could not begin trading.
Meanwhile, in Chicago, the futures market was already in free fall. S&P 500 futures were dropping faster — and earlier — than the underlying stocks in New York. The price gap between futures and cash equities widened. In a normal market, arbitrageurs would narrow that gap — buying cheap futures and selling expensive stocks, or vice versa. But this morning, with half the cash market unable to open, arbitrage was impossible. The price-discovery mechanism had broken down. Chicago and New York were narrating two different realities.
10:00 a.m. The Dow had already fallen more than 100 points. The murmur on the floor had turned to a roar. Traders screamed their orders until their voices cracked, but nearly all of them were sell orders. "A roar like a beehive" was how people who were there later recalled it. So loud you could not hear the person standing next to you. The twenty-two-meter ceiling reflected the sound, turning the entire floor into a single enormous echo chamber. But most of that noise pointed in one direction — selling. Hand signals for buying were almost nowhere to be seen.
11:00 a.m. The DOT system buckled under the load. DOT was the NYSE's electronic order-routing system, the conduit through which brokerage orders reached the specialists on the floor. Introduced in 1976 and upgraded in 1984, it could electronically process orders of up to 100,000 shares. By modern standards, it was primitive. Orders were printed out and delivered to specialists on paper. The dot-matrix printers stationed beside the trading posts could not keep pace with the flood of incoming orders. Paper poured out without end. The pins striking the paper produced an unrelenting zzzzzt, zzzzzt. Torn sheets piled up around the printers; sell-order slips on the floor accumulated ankle-deep. The area around each trading post became a sea of paper. Some printers overheated and jammed, half-printed order tickets dangling in midair. Clerks were seen crawling on the floor to gather the slips and hand them to specialists.
The ticker delay grew from fifteen minutes to thirty, from thirty to sixty. At its worst, it reached seventy-five minutes. The prices on screen were an hour old. No one knew the current price. You could pick up the phone and ask your broker, "What's it trading at now?" — and the broker didn't know either. Orders were submitted, and for over an hour it was impossible to confirm whether they had been executed. A fog of war had descended on the financial markets. Some traders, unable to determine whether their sell orders had been filled, submitted the same orders again. These duplicate orders compounded the selling pressure beyond its actual magnitude.
12:00 p.m. Chairman Phelan walked the floor in person, rallying the specialists. The steady gait of a former Marine. His presence alone was the message: "The NYSE has this under control." A brief rally materialized. From a 200-point loss, the market appeared to recover toward a 120-point decline. A murmur of hope circulated among the traders: "Have we hit bottom?" It did not last. The rally survived less than thirty minutes.
During the lunch hour, a rumor spread. "The SEC may shut down the market." Its source was unclear, but its path of transmission was not. From floor trader to broker, broker to institutional investor, institutional investor to corporate treasurer. The rumor doubled the fear. The logic was simple: sell before the market closes. You cannot sell in a closed market. At 1:00 p.m., the dumping resumed.
2:00 p.m. The Dow fell below the 300-point loss mark. Specialists' capital was running dry. They were obligated to buy shares with their own money to maintain market equilibrium, but the more they bought, the lower prices fell. Each purchase deepened their losses. Duty and survival were colliding. Some specialist firms lost the bulk of their capital in that single day.
3:00 p.m. The final ninety minutes were free fall.
Bids vanished. Even blue-chip stocks like IBM, GE, and Exxon had no buyers, and their prices dropped vertically. IBM plummeted from its previous close of $135 to an intraday low of $104 — a 23 percent decline in one day. At the time, circuit breakers — automatic mechanisms that halt trading when an index drops beyond a certain threshold — did not exist. A car without brakes was hurtling toward a cliff. In some stocks, specialists were unable to post any bid at all. A stock without a buyer has no price — price is the result of a transaction, and without a transaction, it is merely a number.
4:00 p.m. The closing bell rang. Chairman Phelan pressed it himself.
The Dow's opening level: 2,247.06. The close: 1,738.74. Intraday low: 1,677. The decline: 508 points. The percentage drop: 22.61 percent. Greater than Black Thursday (12.8 percent) and Black Tuesday (11.7 percent) of the 1929 crash combined. In a single day.
604 million shares changed hands. Three and a half times the normal volume. Due to system limitations, many trades were not confirmed until hours after the close. In one day, approximately $500 billion — more than 10 percent of U.S. GDP — had evaporated.
When the closing bell rang, what settled over the floor was not cheering but silence — not relief, but a stunned, hollow stillness. Men in expensive suits sat slumped on the floor. Others leaned against trading posts, ties loosened and hanging. Those staring up at the quote boards had vacant eyes. No one made eye contact. The floor was carpeted with the paper that printers had disgorged all day — executed orders, unexecuted orders, half-printed orders tangled together. It would take cleaners hours to sweep it all up. Beneath the twenty-two-meter ceiling, the cathedral of capitalism lay buried in debris.
3
Why did this happen?
The prior week's decline, the cascading crashes across Asia and Europe, a House bill targeting M&A tax provisions, the trade deficit announcement — many causes were cited. But to explain an unprecedented single-day decline of 22.6 percent, macroeconomic context alone is insufficient. The economic fundamentals were unfavorable, but not 22.6-percent unfavorable. The U.S. economy in 1987 was growing. Unemployment was declining. Corporate earnings were solid. Nothing in the economy justified a 22 percent drop in one day.
The conclusion of the Brady Commission — the presidential task force appointed by Ronald Reagan to investigate the crash — was clear. What amplified the scale of the collapse was portfolio insurance.
Portfolio insurance was a direct descendant of the Black-Scholes formula discussed in Chapter 4. The strategy traces its origins to 1979, when Hayne Leland, a finance professor at UC Berkeley, found inspiration at a dinner party. A pension fund manager had said, "I want to enjoy the upside when stocks rise, but I want protection when they fall." Leland thought: that is identical to the payoff structure of a put option. If so, couldn't he replicate that payoff through the reverse engineering of the Black-Scholes formula, without actually buying a put?
Leland refined the idea with his colleague Mark Rubinstein, an expert in options theory with deep practical knowledge of the Black-Scholes model. In 1980, the two joined forces with John O'Brien, a finance industry veteran, to found LOR Associates — the name derived from the first letters of their surnames.
The core idea was simple: when a stock portfolio declines, automatically sell futures to offset the losses. It works like insurance. When stocks go up, you capture the gains; when they go down, futures sales cushion the blow. "You can protect against a downturn without selling a single share" — it was an irresistible promise. Tell a pension fund's board of directors that they can preserve principal while still participating in a rising stock market, and there is no reason to say no.
The technical mechanics worked like this. Using the pricing formula for a put option (a derivative that pays off when stock prices fall) from the Black-Scholes model, the strategy replicates the put's payoff structure through futures trades. When stock prices decline, a computer calculates: "Reduce equity exposure and increase cash." Because selling actual stocks is costly and slow, S&P 500 futures are sold instead. This is called dynamic hedging. "Dynamic" means the hedge ratio is continuously adjusted in response to market conditions. If stocks fall 1 percent, sell a small amount of futures. If they fall 2 percent, sell more. If they fall 5 percent, sell in bulk. The greater the decline, the greater the volume of selling.
Consider a ball rolling down a slope. The faster the ball accelerates, the harder you must press the brake. Portfolio insurance automated that brake. But the analogy is missing something: what if hundreds of balls are rolling down the same slope, in the same direction, all pressing the same brake? What if the act of braking itself makes the slope steeper?
By October 1987, roughly $60 billion to $100 billion in institutional assets were managed under portfolio insurance programs. LOR alone had $50 billion in assets under protection. That simultaneously meant $50 billion worth of automated sell programs would activate the moment the market declined.
Here lay the fatal blind spot. The model assumed that market liquidity was infinite. It presupposed that you could sell futures at any time, in any quantity, at the prevailing price. This was a prerequisite demanded by the mathematics of Black-Scholes — continuous trading, frictionless markets, infinite liquidity. Mathematically necessary assumptions. Assumptions that did not hold in reality.
But tens of billions of dollars were all running the same algorithm. When the market falls 1 percent, these algorithms simultaneously flood the market with futures sell orders. The deluge of sell orders drives futures prices lower. Lower futures prices trigger arbitrageurs to sell cash equities. Falling cash equities cause the portfolio insurance algorithms to compute: "Sell more." A feedback loop in which selling begets selling. Decline breeds selling, selling breeds decline, and decline breeds even more selling.
The insurance had caused the accident.
Fire insurance pays out when a fire occurs, but buying fire insurance does not start one. Auto insurance activates after an accident, but filing a claim does not cause one. Portfolio insurance was different. The very act of all policyholders filing claims simultaneously was what ignited the blaze. Insurance not preventing the disaster, but becoming its cause — a paradox.
On October 19 itself, portfolio insurance programs unleashed approximately $6 billion in futures sell orders. That represented 15 to 20 percent of total futures selling for the day. The concentrated morning selling accelerated the market's decline, and in the afternoon, as programs dumped orders they had been unable to execute that morning, they triggered the final ninety minutes of unbroken collapse. In a market with no buyers, this volume pushed prices off a cliff.
The world that the Black-Scholes formula had assumed — a world in which markets move continuously, liquidity is ample, and prices follow a normal distribution — did not exist that day. Prices did not move in orderly increments — "100 yesterday, 99 today." They jumped from 100 to 78. When the world the formula presupposed shattered, the strategy derived from that formula amplified the shattering.
The formula had destroyed itself. It was the dark side of MacKenzie's concept of "performativity" discussed in Chapter 4. If a formula can make a market, a formula can also unmake one.
4
The next morning, Greenspan issued a single sentence.
Alan Greenspan had been confirmed by the Senate and sworn in as Federal Reserve Chairman in August 1987. A New York native with an unusual background — he had attended the Juilliard School of Music before switching to economics. He had been a devotee of Ayn Rand's laissez-faire philosophy and had served as chairman of the Council of Economic Advisers under President Ford. He was known for his skill at taming inflation and his mastery of opaque language. Two months and eight days into his tenure, he faced the worst single day in financial history. His first test as Fed chairman turned out to be the largest one it had ever produced.
On the 19th itself, he refrained from immediate intervention, choosing to monitor the situation. He needed to determine which direction the storm was moving.
On Tuesday morning, the 20th, a brief statement was released before the market opened:
"The Federal Reserve, consistent with its responsibilities as the Nation's central bank, affirmed today its readiness to serve as a source of liquidity to support the economic and financial system."
One sentence. But that single sentence prevented systemic collapse. Banks were poised to cut off lending to brokerage firms. Given the scale of the previous day's losses, the solvency of those firms was in doubt. Brokerages needed to borrow from banks to meet their customers' margin calls; if banks tightened the spigot, the brokerages would go under. If brokerages went under, settlement failures would cascade like dominoes. Gerald Corrigan, president of the New York Fed, had been calling the CEOs of major banks since the previous night, pressuring them to maintain their credit lines to the securities firms. "If you cut off lending, the brokerages collapse. If the brokerages collapse, you won't recover your loans either. The Fed stands behind you." Corrigan's persuasion was equal parts raw threat and implicit guarantee. Greenspan's statement lent it official authority.
The market rallied. On October 20, the Dow gained 102 points — a 5.9 percent rise. But even that day had its razor-edge moments. S&P 500 futures were down 12 percent when the Chicago Mercantile Exchange halted trading, and on the NYSE, trading was suspended in several stocks, bringing the market to the brink of closure. Goldman Sachs and Salomon Brothers stepped in with large-scale purchases, sparking the recovery. The sharpest moment of the crisis had passed.
But the paradox created by the post-crisis response was more significant than Black Monday's crash itself.
The Brady Commission concluded that portfolio insurance and program trading had amplified the collapse. The report contained key recommendations. First, the introduction of circuit breakers — mechanisms that automatically halt trading when an index drops beyond a specified threshold. In October 1988, the NYSE officially adopted circuit breakers: a one-hour halt if the Dow fell 250 points, a two-hour halt at 400 points. The rationale was "give humans time to think." Since human judgment could not keep pace with algorithmic speed, the solution was to force a stop.
But market participants drew the opposite lesson.
The DOT system delays, the overheated printers, the jammed phone lines — all of it demonstrated the limitations of human beings and analog systems. Submitting an order and not knowing for an hour whether it was filled. A ticker displaying the past instead of the present. Placing an order by phone, only to find the line dead. Specialists buying beyond their capital limits and still unable to absorb the volume.
Regulators warned that "algorithms are dangerous." Market participants concluded that "analog is more dangerous." The problem, as they saw it, was not that algorithms ran too fast but that the infrastructure could not keep up. After Black Monday, Wall Street did not scale back its electronic trading infrastructure — it expanded it dramatically. Faster systems, greater capacity, more sophisticated algorithms. While the fire chief was saying "fire is dangerous," the architects were building bigger furnaces. The larger the furnace, the greater its efficiency — and the greater the devastation when the fire escapes its confines.
Greenspan's single sentence left its own paradox. It established a precedent: "The Fed will intervene in a crisis." This would later be called the "Greenspan Put" — the expectation that the Fed would come to the rescue whenever markets fell sharply. A seed of moral hazard had been planted, encouraging market participants to take ever-greater risks. Insurance makes humans bolder — it is human nature. Buckle your seatbelt, and you drive faster. Wear a life jacket, and you swim into deeper water.
Twenty-three years later, on May 6, 2010, the Dow Jones plummeted roughly 998 points in thirty-six minutes beginning at 2:32 p.m., then rebounded. At its intraday low, the index touched 9,872 — down 9.2 percent from the previous close. The Flash Crash. The kind of collapse that in 1987 took all day on a floor of screaming humans occurred in 2010 in a matter of minutes, with no time for human intervention. The analog market of 1987, with its printers spewing paper, had vanished, replaced by fiber-optic cables and microsecond algorithms. Yet the result was the same — a price collapse, the evaporation of liquidity, and human helplessness. The pattern of an algorithmic market — "efficient most of the time, catastrophic once in a while" — had begun in 1987 and was only accelerating.
5
Before the ashes of Black Monday had even cooled, one man was already building a new game.
John Meriwether. The legendary head of the bond arbitrage desk at Salomon Brothers. Described as "the emperor of Salomon" in Michael Lewis's Liar's Poker.
Meriwether grew up in Roseland, on Chicago's South Side, in an Irish Catholic family. After earning his MBA at Northwestern University, he joined Salomon Brothers in 1973. Assigned to the bond trading desk, he was running the arbitrage group within four years, by 1977. His team's share of Salomon's total revenue fluctuated between 80 and 100 percent. This single team generated more profit than every other division in the firm combined.
Lewis's portrait of Meriwether in Liar's Poker was of a quiet but lethal man. When Salomon chairman John Gutfreund challenged Meriwether to a game of Liar's Poker for $10 million, Meriwether made an audacious counteroffer — signaling that his style was to make money quietly in the back room, not to gamble. But the stakes of the games he actually played dwarfed Gutfreund's wager.
In 1991, it was discovered that a subordinate, Paul Mozer, had submitted false bids in U.S. Treasury auctions to acquire holdings exceeding the regulatory limit of 35 percent of any issue. Meriwether had not personally participated in the manipulation, but he bore responsibility for failing to immediately report Mozer's actions to the authorities after being informed. He resigned alongside CEO Gutfreund and President Thomas Strauss. He paid a $50,000 fine to the SEC, accepted a three-month trading suspension, and left Salomon.
A man who has walked away has two choices. Retire, or start something bigger.
On February 24, 1994, Meriwether executed the first trade of his new fund from Greenwich, Connecticut. Long-Term Capital Management — LTCM. Initial capital: $1.01 billion. The name alone radiated confidence. "Long-Term Capital Management" — a declaration of intent to manage capital over the long run.
Greenwich. This coastal Connecticut town, a forty-minute Metro-North ride from Manhattan, was becoming the hedge fund capital of the world through the 1990s. Steamboat Road had earned the nickname "Hedge Row" for the density of fund offices clustered there. LTCM's office was in a building called Osprey House, near the harbor. Annual rent: $1.5 million. Not a glass tower in Manhattan but a quiet, low-rise building surrounded by trees. A lawn and a parking lot out front; inside, along quiet corridors, the sounds of ringing phones on the trading desk and Bloomberg terminal alerts. A space designed for focus on mathematics and models, away from Wall Street's noise. By 1996, the staff had grown to 150, with additional offices in London and Tokyo.
The roster of partners Meriwether assembled was the most luminous in Wall Street history. Robert Merton and Myron Scholes — co-creators of the Black-Scholes option pricing model discussed in Chapter 4. In 1997, the two would receive the Nobel Prize in Economics. They received it while serving as LTCM partners. As the Nobel committee lauded them for "developing a new method for determining the value of derivatives," the limitations of that method were being tested in reality. David Mullins, former vice chairman of the Federal Reserve Board, also joined. So did Eric Rosenfeld, Larry Hilibrand, Victor Haghani, and other elite bond arbitrageurs who had worked under Meriwether at Salomon. The finest minds in academia, connections to the regulatory establishment, and battle-tested traders — all gathered under one roof.
Raising capital was not the hard part — the hard part was turning investors away. The minimum investment was $10 million. A three-year lock-up with no redemptions. Fees were 2 percent management plus 25 percent of profits — higher than the industry-standard "2 and 20." And still, the Bank of Italy, the Bank of China, the Kuwait pension fund, and university endowments from around the world queued up. Being able to invest in a fund run by Nobel laureates was, in itself, a badge of status on Wall Street.
LTCM's strategy was called convergence arbitrage. The core logic went like this: when two assets that should be mathematically similar are temporarily mispriced, sell the expensive one and buy the cheap one. In time, their prices converge, yielding profits on both sides.
For example, among U.S. Treasuries, there are "on-the-run" bonds (the most recently issued) and "off-the-run" bonds (older issues). Both carry the full faith and credit of the U.S. government, so their credit risk is identical. Yet on-the-run Treasuries trade more actively and are therefore slightly more expensive. This price difference — according to LTCM's models — was certain to narrow. Time would take care of it.
The problem was that this price difference was extremely small. To extract meaningful profits from a spread of less than 0.1 percent, you had to bet on an enormous scale. LTCM used leverage — borrowing twenty-five to thirty times its equity. Multiply a 0.1 percent spread by 25x leverage, and you get a 2.5 percent return on equity. Run dozens of such trades simultaneously, and the annual return reaches 40 percent. Mathematically, it was flawless.
The early results were spectacular.
1994 (ten months of the first year): approximately 20 percent. 1995: 43 percent. 1996: 41 percent — all net of fees. Compared to the S&P 500's 20-to-30 percent annual returns over the same period, LTCM's performance was dominant, and its volatility far lower. It earned the moniker "the fund of geniuses." Major Wall Street banks considered it a privilege to do business with LTCM. Goldman Sachs, JPMorgan, Merrill Lynch, Deutsche Bank, UBS — these banks provided LTCM with leverage while remaining largely in the dark about the fund's specific positions. They trusted the geniuses.
In 1997, returns fell to 17 percent. The S&P 500 returned 33 percent. Other funds that had copied LTCM's strategy were chasing the same opportunities, intensifying competition. Arbitrage spreads were narrowing. The pool of mathematically extractable profit was shrinking, while the number of funds pursuing it was growing.
Meriwether chose a paradoxical solution. At the end of 1997, he forcibly returned $2.7 billion in capital to investors. He did not reduce the fund's total assets — instead, he reduced its equity base, pushing the leverage ratio even higher. Less principal, bigger bets. When the vein of ore narrows, you drill deeper. Investors objected, but Meriwether's answer was blunt: lower equity means higher returns. That was the math.
By early 1998, LTCM's equity stood at approximately $4.7 billion. Total assets exceeded $125 billion. The leverage ratio was roughly 27 to 1. And the notional value of derivatives — the face amount written into contracts, though no actual cash changes hands for most of it — reached approximately $1.25 trillion. Roughly 15 percent of U.S. GDP in 1998 was being managed from a quiet office in Greenwich. While yachts bobbed peacefully in the harbor visible beyond the lawn, inside that building, contracts representing 15 percent of the world economy scrolled across Bloomberg terminal screens.
6
On August 17, 1998, Russia suspended payments on its sovereign debt.
The Russian government abandoned its defense of the ruble and declared a moratorium — a suspension of debt payments — on its ruble-denominated bonds (GKOs). The ruble collapsed. The exchange rate, which had stood at roughly 6.3 rubles per dollar before August 17, soared to 21 rubles per dollar after the central bank shifted to a fully floating exchange rate on September 2. In three months, the currency had lost two-thirds of its value.
The aftershocks of the Asian financial crisis had reached Russia. From the collapse of the Thai baht in July 1997, the crisis had taken roughly a year to travel through Indonesia and South Korea before arriving in Moscow. Asia's economic slowdown had depressed demand for crude oil and other commodities. Oil prices plunged more than 33 percent between 1997 and 1998. With oil and natural gas — the backbone of Russian exports — crumbling in price, an already fragile fiscal position became untenable.
To understand why one emerging-market country's debt default brought down a hedge fund in Connecticut, you have to look at the deepest assumption embedded in LTCM's models.
LTCM's models operated on the basis of correlation — the statistical relationship between asset prices. "When A rises, B rises too" — the fund constructed its positions by exploiting such historical patterns. Risky assets move with risky assets; safe assets move with safe assets. As long as these correlations held, LTCM's positions hedged one another. When one side lost money, the other side made money. A loss on Italian government bonds would be offset by a gain on German Bunds. A loss on Danish mortgage bonds would be neutralized by profits on swap positions. Dozens of positions formed an intricate lattice, each supporting the others.
The Russian moratorium blew those correlations apart.
Panic-stricken investors sold anything that carried risk. Not just Russian bonds — Brazilian bonds, Italian bonds, mortgage-backed securities, corporate bonds, emerging-market currencies — all of it, indiscriminately. And all of that money flowed to a single destination: U.S. Treasuries. This is called "flight to quality." Every risky asset falling in unison; every safe asset rising in unison. A one-way street. Assets that normally moved in different directions were, in the face of panic, all moving the same way. Correlations had suddenly converged to one. The magic of diversification had vanished.
The spreads that LTCM had bet would narrow widened at a pace without historical precedent. Prices that the fund had believed would converge were instead diverging. A bridge engineered to span a narrow river had, overnight, been set down across an ocean.
In August alone, LTCM lost $1.8 billion — 44 percent of its capital. From the start of the year, $2.5 billion had evaporated — 52 percent of its capital.
On September 2, LTCM sent a letter to its investors disclosing the August losses. When this letter circulated through the market, LTCM's predicament became publicly known. And once LTCM's predicament was known, other funds holding the same positions scrambled to get out first. A race to claim lifeboats on a sinking ship. Assets of the same type that LTCM held were dumped on the market, widening spreads further. The larger LTCM's losses grew, the greater the exit pressure from other funds; and the greater the exit pressure, the larger LTCM's losses became. The feedback loop of Black Monday was repeating itself, this time in the bond market.
By September, $500 million was evaporating each week. On Monday, September 21, alone, $553 million vanished — the worst single day in LTCM's history, five times the maximum daily loss the models had predicted. Equity collapsed below $600 million. Assets of $125 billion were now propped up by $600 million in capital. The leverage ratio had surged past 200 to 1.
In the Greenwich office, Bloomberg terminal screens glowed red. Two Nobel laureates watched as their models diverged from reality. In the world of Merton and Scholes's formulas, the probability of such an event was less than once in a hundred years. But reality does not defer to a formula's probabilities. Lunch breaks grew shorter; the office lights stayed on deep into the night. Only the sound of lawn sprinklers broke the silence before dawn.
What would happen if LTCM collapsed? Its counterparties included more than fourteen major financial institutions. If LTCM went bankrupt, $1.25 trillion in derivative contracts would be forcibly liquidated. If that volume hit the market, every financial institution's positions would sustain cascading damage. If Goldman Sachs's swap contracts with LTCM were voided, Goldman would suffer losses on the other positions it had been hedging through those contracts. If Goldman sold assets to cover those losses, other banks holding the same assets would take hits too. Once the first domino falls, there is no stopping. All of Wall Street was in the same boat, and the boat was sinking.
On September 18, LTCM's management contacted William McDonough, president of the Federal Reserve Bank of New York. Two days later, on September 20, a New York Fed team visited LTCM's offices in Greenwich. When they opened the books, the situation was far worse than the outside world realized.
Wednesday, September 23. Tenth floor, Federal Reserve Bank of New York. 33 Liberty Street. In a conference room with views of the Manhattan skyline, Wall Street's chieftains assembled. Goldman Sachs, Merrill Lynch, Morgan Stanley, JPMorgan, UBS, Deutsche Bank, Barclays, Credit Suisse. Men in dark suits and ties sat around a long table. McDonough's message was simple: either save LTCM or watch the financial system collapse. The choice is yours, but the consequences belong to everyone.
By around 6:00 p.m. that evening, an agreement was reached. Fourteen financial institutions contributed a total of $3.625 billion. Ninety percent of LTCM's equity was transferred to the consortium of creditors. Meriwether and the partners lost everything. Most of the personal fortunes of two Nobel laureates evaporated along with it.
Bear Stearns alone refused to participate. Despite being one of LTCM's primary clearing brokers, it declined to contribute additional capital, claiming, "We already carry enough risk." It had already exercised roughly $500 million in claims against LTCM. The refusal was long remembered on Wall Street. The bank that turned its back when solidarity was needed. Exactly ten years later, in March 2008, Bear Stearns was the first to fall in the subprime crisis. It was acquired by JPMorgan at $2 per share — 99 percent below its 52-week high of $170. The bank that had refused to help others in 1998 could find no bank willing to help it in 2008. Wall Street has a long memory.
No public funds were injected. The solution was financed entirely with private-sector money. But the private banks contributed not voluntarily — they had been summoned to the Fed's conference room and told, "You will go down with it." The debate over moral hazard — "privatize the gains, socialize the losses" — was seeded here. The seed that had germinated with Greenspan's single sentence in 1987 grew with the LTCM rescue in 1998 and became a towering tree with the TARP bailouts of the 2008 financial crisis.
7
Black Monday and LTCM. Separated by eleven years, the two events posed the same question at different scales.
The parallels are stark. In both cases, the assumptions underlying mathematical models broke down under extreme conditions. Portfolio insurance assumed that "markets always have liquidity." LTCM assumed that "correlations between asset prices hold." Under normal market conditions, these assumptions are largely correct. Ninety-nine percent of the time, the models work. The problem is the other 1 percent.
In a normal distribution — the mathematical foundation of the Black-Scholes model — the probability of an event more than four standard deviations from the mean is roughly 6 in 100,000. If each trading day is one observation, that is approximately once every forty-five years. Five standard deviations: once in approximately seven million years. Six standard deviations: effectively never.
But what happens in real financial markets? Black Monday's 22.6 percent decline was an event of more than twenty-five standard deviations. Under the assumption of a normal distribution, even if the market opened every day for longer than the age of the universe (13.8 billion years), this should not have occurred. Yet it did. LTCM's losses were no different. The scenario that the models had computed as "once in a million years" materialized in a single month.
The problem lay in the normal distribution itself. The normal distribution is adequate for predicting the weather in Seoul. Most days fall between minus 10 and plus 35 degrees Celsius; minus 20 or plus 40 is extremely rare. But financial markets are not like Seoul's weather — they are more like earthquakes around the world. Most days register no tremor, but when a quake strikes, it might measure 7 on the Richter scale, or 9. Small tremors accumulate into large quakes, and the aftershocks of large quakes trigger still more. Financial markets work the same way. When feedback loops of fear take hold, the magnitude of the swings exceeds anything the mathematics predicted.
Real-world financial data does not follow a normal distribution. Extreme events — tail risk — occur far more frequently than the bell curve predicts. The far ends, or "tails," are thicker than the model implies. In statistics, this is called a "fat tail." It is the mathematical reality behind what Nassim Taleb would later term the "Black Swan."
In Chapter 4, the Black-Scholes formula asked: "What is the price of uncertainty?" The answer — a single number called volatility — opened the options market, created the derivatives universe, and permanently changed the language of finance.
Black Monday and LTCM posed a different question: is the very confidence that uncertainty can be priced itself uncertain?
The pattern Newton displayed during the South Sea Bubble repeated itself. Newton was the genius who discovered the law of universal gravitation, yet he could not calculate human madness. "I can calculate the motions of heavenly bodies, but not the madness of people" — his lament echoed 270 years later in a Greenwich office. Merton and Scholes were Nobel laureates, but they failed to capture the market's extremes within their models. The tools had changed. Bills of exchange had become derivatives; intuition had become differential equations. But before the question "Can humans (or human-made models) achieve perfect rationality?" the answer remained the same as it had been six hundred years ago. No.
LTCM's collapse should have left the market with a warning. A model built by two Nobel laureates had crumbled before reality — that should have provoked a fundamental doubt about mathematical certainty. But Wall Street extracted a different lesson: "The model wasn't wrong — it just wasn't sophisticated enough."
LTCM had concentrated on bond arbitrage and collapsed when correlations broke. The takeaway, as Wall Street saw it: bundle more assets, diversify more broadly, use more sophisticated correlation models. The ultimate expression of this logic was the CDO — the collateralized debt obligation.
More complex models, more layers, more elaborate structures — the CDO was a monument to that ambition. Thousands of mortgages bundled together, the bundle sliced into tranches and rated, and those slices bundled again and rated once more. CDOs of CDOs — so-called "CDO-squareds" — were created. The number of human beings on Earth who fully understood these structures was vanishingly small. That fact alone was a warning sign, but no one stopped. The lesson of LTCM should have been "be humble." The lesson that was extracted was "be more sophisticated." And that sophistication led to the subprime crisis of 2008.
In Chapter 6, that story continues.