The fund with Wall Street's most prestigious board
John Meriwether, a legendary fixed-income trader who had run Salomon Brothers' arbitrage desk, founded Long-Term Capital Management in 1994 and assembled, literally, the best academic and professional team money could buy. Sitting on its board were Myron Scholes and Robert Merton, who in 1997 would receive the Nobel Prize in Economics precisely for the options-pricing model (Black-Scholes-Merton) that underpinned much of the fund's own strategy.
The team's reputation was so strong that LTCM attracted capital from some of the world's largest banks and institutional investors, and negotiated exceptionally favorable financing terms precisely because no one doubted its technical sophistication.
The strategy: betting on convergence, with extreme leverage
LTCM specialized in so-called 'convergence arbitrage': identifying pairs of assets with a theoretically related price (for example, two very similar Treasury bonds) whose market price diverged slightly for technical or liquidity reasons, and betting that the small gap would close over time, exactly as its statistical models, built on historical data, predicted.
The problem with these opportunities is that the price differences are usually tiny, which pushes any fund chasing them to use extreme leverage to make the profit meaningful. LTCM came to control positions with a notional value exceeding $1 trillion with only about $5 billion in its own capital: leverage that, at its most aggressive, topped 25 to 1.
- Strategy: convergence arbitrage between fixed-income assets with theoretically related prices.
- Leverage: over 25 times its own capital in its most aggressive positions.
- Size: positions with a notional value exceeding $1 trillion.
- Model basis: historical correlations between assets calculated from past data.
August 1998: when Russia broke the model's correlations
In August 1998, Russia declared a moratorium on its sovereign debt and devalued the ruble, triggering what investors call a 'flight to quality': everyone tried to sell risk assets at the same time to shelter in the safest possible bonds, mainly US and German debt.
LTCM's models assumed certain assets would move in relatively predictable ways relative to each other, based on decades of historical data. But in a widespread panic, almost every investor in the world wants to sell risk and buy safety at the same time, regardless of the technical differences between assets that, under normal circumstances, would determine their relative price. The price gaps LTCM expected to narrow widened instead, and they did so simultaneously across dozens of different positions that, according to the model, shouldn't have been so correlated with each other.
$4.6 billion lost and a Fed-orchestrated bailout
Between May and September 1998, LTCM lost approximately $4.6 billion, close to 90% of its capital, at a pace that in some weeks exceeded $500 million a day. Given the size and interconnection of its positions with the world's major financial institutions, there was a well-founded fear that a forced, disorderly liquidation of those positions could destabilize the entire global financial system, not just the fund itself.
The Federal Reserve Bank of New York, without directly contributing public money, organized and pressured a consortium of fourteen major banks into jointly injecting $3.6 billion in exchange for effective control of the fund, allowing its positions to be unwound in an orderly manner over several months instead of triggering a simultaneous panic sale. The fund was fully liquidated in early 2000.
What LTCM teaches us about model risk and leverage
LTCM's collapse became the go-to case study on 'model risk': no mathematical model, however sophisticated, can fully capture how markets will behave during a truly extreme event, precisely because such events are, almost by definition, different from everything in the historical data the model was trained on.
It also illustrates the specific danger of extreme leverage, the same theme that reappears in the Great Depression and the 2008 crisis: when you control a position twenty-five times larger than your capital, you don't even need to be wrong about where prices will go long-term. It's enough for the market to move against you long enough, and violently enough, that margin calls force you to liquidate before your thesis ever gets the chance to play out. The team's intelligence was never in question; what failed was assuming the past limited what the future could do.
Apply This Knowledge to Your Real Finances
eXcenda combines net worth, expenses, simulations, and macro context into a single app so you can move from theory to action.
Frequently Asked Questions
Myron Scholes and Robert Merton, awarded the Nobel Prize in Economics in 1997 for their work on options pricing (the Black-Scholes-Merton model), sat on the firm's board.
Not directly. The Federal Reserve Bank of New York organized and coordinated the operation, but the $3.6 billion bailout capital came from a consortium of fourteen private banks in exchange for taking control of the fund.
In its most aggressive positions, the fund came to control assets worth more than twenty-five times its own capital, and its total notional value in derivatives exceeded $1 trillion at one point.
Because its positions were intertwined with dozens of major banks and counterparties around the world. A forced, disorderly liquidation could have triggered cascading losses at those same institutions, with the risk of contagion spreading through the whole system.
It's the risk that a mathematical model, even if well built and validated against historical data, fails when facing a genuinely extreme scenario that isn't adequately represented in that past data.