Long Term Capital Management
Long-Term Capital Management (LTCM) was a hedge fund founded in 1994 by John Meriwether, with a board of directors that included Nobel laureates Robert Merton and Myron Scholes. The fund employed sophisticated mathematical models — primarily variants of the Black-Scholes model — to identify and exploit pricing inefficiencies across global markets. For four years, it delivered extraordinary returns with low volatility, seemingly having solved the problem of risk management through quantitative rigor.
In September 1998, LTCM collapsed. The proximate cause was Russia's sovereign debt default, which triggered a flight to quality that LTCM's models had not anticipated. The models assumed that correlations between markets were stable and that historical relationships would persist. When correlations spiked — when previously uncorrelated assets moved in lockstep — the fund's highly leveraged positions, which were market-neutral in normal conditions, became catastrophically concentrated. The losses were not merely large; they were systemic. LTCM's counterparties included most major global banks, and its failure threatened to cascade through the financial system. The Federal Reserve organized a private bailout to prevent a broader collapse.
The LTCM collapse is a canonical case study in the limits of mathematical modeling in complex systems. The models were not wrong in their narrow domain; they were wrong in their assumption that the narrow domain was the relevant domain. The risk that destroyed LTCM was not in the models; it was in the model risk — the risk that the model itself was an incomplete representation of the system. The Black Swan that killed LTCM was not a statistical tail event. It was a structural regime change that the model's architecture could not represent.
LTCM's collapse was not a failure of mathematics. It was a failure of epistemology — the belief that a sufficiently precise model of the parts could predict the behavior of the whole. The mathematics was elegant. The world was not.