Cover of When Genius Failed

When Genius Failed

Roger Lowenstein

6 ideas

  1. Models Assume Liquidity That Vanishes In Crisis

    Risk models priced positions as if markets would stay liquid and price moves would follow normal distributions, but in a panic spreads widen, correlations across unrelated assets jump to one, and exits disappear. The very event the model deemed near-impossible becomes the event that destroys the firm.

  2. Nobel Laureates Cannot Outrun A Panic

    LTCM was staffed by Myron Scholes, Robert Merton, and elite traders whose intellectual firepower convinced banks to lend almost without limit. Their reputations themselves became a hazard, because counterparties stopped scrutinizing the leverage and concentration that ultimately required a Fed-orchestrated rescue.

  3. Leverage converts small edges into ruin

    LTCM earned thin spreads on convergence trades and multiplied them with leverage near 25-to-1 on the balance sheet and far more through derivatives. That magnification makes a strategy with positive expected value fatal, because a temporary adverse move can wipe out the capital before prices converge. Being right in the long run is worthless if you cannot survive the path to it.

  4. Crowded trades and the liquidity trap

    LTCM's success led rival banks to copy its positions, so the same trades were held by many firms at once. When LTCM needed to exit, it was too large relative to the market and its counterparties were selling the same things or trading against it. The market's liquidity disappeared exactly when the fund needed it.

  5. Fed-brokered bailout to avert systemic collapse

    In September 1998, the New York Fed convened fourteen Wall Street banks, which put up about $3.6 billion to take over LTCM rather than let a disorderly default ripple through its enormous web of counterparties. No public money was spent, but the intervention showed that a private fund could become too interconnected to fail. It raised the moral-hazard concern that sophisticated players would expect rescue in the future.

  6. Correlations converge to one in crises

    LTCM's risk models treated its hundreds of trades across countries and markets as independent bets because their historical correlations were low. In a panic, however, investors dump whatever is risky and illiquid at the same time. Diversification that looks real in calm data then vanishes, and positions that seemed unrelated all lose together.

Save and mark ideas in the app