John Meriwether did something different in the fall of 1987, when senior executives at Salomon Brothers were making rash, gut-driven bets worth billions of dollars while the stock market was collapsing all around him. He may have been twenty yards away. He observed. He then made a calculation. Then he instructed the young quantitative analysts he had hired from graduate finance programs to choose their two or three most promising trades and triple them.
According to one observer, these men’s bodies appeared to function primarily as life support systems for their brains. They made a $50 million profit when they cashed out their Treasury bond holdings three weeks after the crash, when the markets had calmed and the panic had subsided. The lesson Meriwether learned from that encounter was straightforward and, looking back, deadly alluring: those who were prepared to think clearly could take advantage of other people’s fear.
| Fund Name | Long-Term Capital Management L.P. (LTCM) |
|---|---|
| Founded | 1994 |
| Dissolved | 2000 |
| Founder | John W. Meriwether (former Vice-Chairman, Salomon Brothers) |
| Headquarters | Greenwich, Connecticut, USA |
| Notable Board Members | Myron Scholes & Robert C. Merton (Nobel Prize in Economics, 1997) |
| Initial Capital | $1.01 billion (at launch, February 24, 1994) |
| Peak Assets | ~$129 billion (on $4.7 billion equity — leverage ratio 25:1) |
| Off-Balance Sheet Derivatives | ~$1.25 trillion notional value |
| Peak Annual Returns | 43% (Year 2); 41% (Year 3) — after fees |
| Total Losses (1998) | $4.6 billion in less than four months |
| Single Worst Day | August 21, 1998 — lost $550 million in one session |
| Bailout | $3.65 billion from 14 banks, brokered by Federal Reserve Bank of New York (September 23, 1998) |
| Reference Website | The Guardian |
This idea served as the cornerstone of Long-Term Capital Management, the hedge fund Meriwether founded in Greenwich, Connecticut, in 1994 after leaving Salomon Brothers under a cloud due to a Treasury bond scandal that ended his tenure there even though it wasn’t his fault. Wall Street had never seen a company like the one he founded.
People who had truly reinvented finance mathematics—not just applied it, but expanded it—worked there. Two of LTCM’s directors, Myron Scholes and Robert Merton, shared the 1997 Nobel Prize in Economics for creating the Black-Scholes model, which serves as the foundation for financial derivatives pricing in the majority of the world. It is uncommon to have Nobel laureates on your investment committee. It’s quite different to have them three years prior to winning the prize.
Finding bond pairs that should, according to mathematical reasoning, trade at a predictable spread in relation to one another was the fundamental strategy’s elegant concept. You could wager that it would eventually converge back when panic or irrationality caused it to spread wider than it should have. On-the-run versus off-the-run Treasury bonds—newly issued 30-year bonds trading at a premium to older 29.75-year bonds that were nearly identical in every significant way—were the most well-known example of this trade. Because the new bonds were easier to resell, more liquid, and traded more frequently, there was a premium.
LTCM would purchase the less expensive older bond, short the more expensive new one, and watch for the premium to decline as the new bond aged. In the early years, it was remarkably consistent. The returns, which were 21% in the first year, 43% in the second, and 41% in the third, are the kind of figures that cause serious institutional investors to forget what they were supposed to be asking.
Leverage was the issue, and it took years for the issue to become fully apparent. The returns on any given trade were minimal since the price differences LTCM was taking advantage of were measured in basis points rather than percentage points. LTCM borrowed heavily in order to produce the absolute dollar profits that the fund’s investors anticipated.
By early 1998, the portfolio had approximately $129 billion in assets on a capital base of $4.7 billion, with a leverage ratio greater than 25 to 1. In addition, the fund had amassed off-balance-sheet derivative positions worth about $1.25 trillion. These weren’t careless rogue positions. They were the logical progression of a safe strategy, according to the models. Regarding the math, the models were correct. There was something else they were lacking.
The behavior of markets under conditions that the models assigned nearly zero probability was what they were missing, as Seth Klarman had argued from the start and Warren Buffett realized when Meriwether approached him in 1993 to invest and he graciously declined. The one in fifty million situation. Salomon Brothers declared on July 17, 1998, that it was closing its own convergence trades, which proved to be almost exactly the same as LTCM’s holdings. Salomon’s selling pressure worked against every trade they made, causing the fund to decline by about 10%.
Russia then fell behind on its debt on August 17. Now, it’s difficult to adequately describe the impact that had on international bond markets. Reason had no effect on the fear that swept through Tokyo, London, and New York’s trading floors. Institutions that had been placing the same kinds of convergence bets as LTCM rushed in unison toward the exits as it reacted to itself, compounding with every hour.
In Meriwether’s own analogy, the red dollar and the blue dollar, which should have been worth the same amount in five years, abruptly changed their values to 25 cents and $3, respectively. The worst day in the brief history of quantitative finance was August 21, 1998. Before the closing bell, LTCM lost $550 million.
It’s worth slowing down for what happened next because it has an almost mythological quality. One of the first “young professors” to follow Meriwether from Salomon, Victor Haghani, captured the feeling in a way that sticks in your memory: it was as though someone out there was holding their exact portfolio, three times larger, and was liquidating it all at once.
That wasn’t just a metaphor. As word got out on Wall Street that LTCM was weakened, other traders started selling the same instruments ahead of the anticipated forced liquidation because they knew what the fund held because LTCM had been forced to distribute its trades across several banks. Because so many people had written insurance against it, the likelihood of the hurricane increased, as Meriwether later put it. To survive long enough for its trades to recover, the fund required $1.5 billion. It was unable to locate it in time.
The Federal Reserve Bank of New York intervened, organizing a group of 14 Wall Street banks to provide a $3.65 billion recapitalization in exchange for 90% of the company, rather than using government funds. Alongside Warren Buffett, Goldman Sachs made a rival offer to purchase the entire business outright, which would have completely destroyed the current partners. LTCM lost an additional $500 million in a single session on September 21, the day that offer was being discussed.
In the end, the consortium agreement was approved. For the first time in fifteen years, Meriwether and his Nobel laureates were paid $250,000, which is the salary of a junior bond trader without a bonus. Perhaps the most painful aspect of the conclusion was the humiliation of that number. Early in 2000, the fund was fully disbanded.
When you watch the LTCM story from a distance, even though it is one of the most well-documented financial collapses in history, you are not struck by the people’s conceit, even though that is the easy reading. In one-on-one conversations, Meriwether and his team were, by most accounts, genuinely humble about the limitations of their models.
Their lack of self-confidence was not the cause of their failure. It was their model’s inability to account for the possibility that being right about the math is no protection when the market decides, temporarily and irrationally, that the math doesn’t matter. This was something Eugene Fama had cautioned about. Paul Samuelson had been concerned about unusual occurrences.
The eventual failure of the fund did not disprove quantitative finance. It demonstrated that no model can adequately account for factors like liquidity, leverage, and the actions of scared institutions; in 1998, this difference was precisely $4.6 billion.
