In the late 1990s, geniuses had cracked the code. The internet was reshaping retail and .com valuations reflected an expected paradigm shift. Companies like Enron had built what looked like perpetual revenue machines, with managers rewarded handsomely. And academic practitioners descending from their ivory towers entered the investment world with arbitrage strategies that were sure to generate billions in trading profits. To question any of these sages was to expose your ignorance to the world. In When Genius Failed, now over 25 years old, Roger Lowenstein tells the story of Long-Term Capital Management, the foolproof strategy that never was. Readers may shy away from the book based on its age, but it’s not just a fascinating read on market history, it’s a warning for market participants in 2026.
LTCM’s strategy was deceptively simple. John Meriwether’s team, including Nobel laureates Myron Scholes and Robert Merton, identified small mispricings between related securities and waited for them to converge. Each trade earned pennies. Leverage of roughly 100 to 1 turned pennies into billions. The Nobel-grade math assumed continuous markets and normally distributed returns, both of which broke when Russia defaulted in August 1998 and counterparties pulled back simultaneously. The fund lost $4.6 billion in five weeks and required a Fed-orchestrated bailout from fourteen Wall Street banks.
The investor lesson isn’t “leverage is dangerous,” which most readers already know. The deeper lesson is that liquidity is path-dependent. LTCM’s trades were correct in the long run, the bank consortium that took over the book in 1998 ultimately made money on the same positions. What broke the fund was the assumption that funding would remain available while waiting for convergence. When repo lenders pulled back and bid-ask spreads widened, the fund couldn’t hold positions long enough to be right. The market can remain irrational longer than you can remain solvent.
Lowenstein writes well and glosses over technical detail, making the book unusually accessible to non-specialists. Its real value in 2026 is structural relevance. Strategies that look infallible while they’re working can unravel quickly when their hidden assumptions break. When Genius Failed remains the best primer available for thinking clearly about that pattern.
