What really happened

Long-Term Capital Management was 1994’s most prestigious hedge-fund launch in Wall Street history. Founder John Meriwether — bond trading legend from Salomon Brothers — assembled a team of the best quants and two Nobel laureates: Myron Scholes (Black-Scholes option pricing model) and Robert Merton. The strategy: relative-value trades in bond markets — buy what looked underpriced, short what looked overpriced, wait for convergence.

The early years were spectacular: 1994-1997 LTCM returned 40% after fees — by mid-1998 the fund had $4.7B in equity. But the trades were often microscopic (spread differences of cents per bond point). To make money on those, you needed leverage. LTCM operated at 25:1 balance-sheet leverage — plus off-balance-sheet derivatives that lifted effective leverage to around 100:1. For every dollar of equity, $100 of market risk.

On August 17, 1998 Russia unilaterally defaulted on its ruble bonds. Global bond spreads exploded. “Convergence trades” diverged massively. LTCM lost $1.85B in 4 weeks — almost half its equity. The banks financing LTCM issued margin calls. A bankruptcy likely would have had systemic consequences — LTCM positions were so large that liquidating them would have destabilized the market. On September 23 the NY Fed organized a privately funded bailout: 14 banks put up $3.6B, took over LTCM and unwound positions in orderly fashion.

The warning signs everyone ignored

The leverage was public knowledge — LTCM had disclosed it in bank negotiations. But most lenders assumed that the Nobel laureates’ quantitative models controlled the risk sufficiently. The fundamental problem: the models were based on the previous 20 years of volatility data. That sample contained no crisis at Russia-1998 magnitude. The models calculated a “6-sigma event” as impossible — and exactly that occurred.

The combination of 100:1 leverage and “models as risk protection” is dangerous. Quantitative risk management only works in stress scenarios the model has seen. In genuinely new crises — so-called tail events — all models fail. LTCM operated on the assumption that historical correlations remain stable. Once global liquidity vanishes, all risk assets correlate — and diversification benefits disappear exactly when you need them most.

What investors can learn today

First: leverage optimization is a tail risk. A strategy that makes 20% with 100:1 leverage needs only a 1% market move to wipe out the entire capital base. Second: quantitative models have blind spots for tail events. What never happened is overlooked in backtests — and is exactly what really hurts. Third: diversification doesn’t work in a crisis. When everyone just wants to sell, all asset classes correlate. Real crisis diversification is liquidity (cash, US Treasuries) — not complexity.