When Genius Failed Ch. 1: The Best Team Ever Assembled — and Why That Is the Point

Investment Background

In 1994 a hedge fund called Long-Term Capital Management began trading. Four years later it had lost $4.6 billion, came close to taking the American financial system down with it, and forced the Federal Reserve to convene fourteen Wall Street banks for an emergency rescue.

Reading that, most people reach instinctively for a conclusion: these people must have been fools, or crooks.

That conclusion is wrong. And if you accept it, this book is worthless to you.

Consider who was actually in the building.

Myron Scholes and Robert Merton, winners of the 1997 Nobel Prize in Economics. The award was for creating, with the late Fischer Black, the theory of option pricing. Every person trading options anywhere in the world today — including every reader of Option Volatility and Pricing in this library — uses their formula. This is not a vague academic honor. It is one of the load-bearing foundations of modern finance.

John Meriwether, founder of the bond arbitrage desk at Salomon Brothers. Under him that desk produced one of the steadiest profit records in Wall Street history through the 1980s. He was widely regarded as the finest bond trader of his generation.

David Mullins, former Vice Chairman of the Federal Reserve. The number two official at the American central bank, recently departed, was sitting at this fund.

Add the elite trading team brought over from Salomon, and a cohort of PhDs from Harvard and MIT.

This was the highest concentration of intelligence ever assembled for the single purpose of making money. That is not in dispute.

The Wall Street Translation

So why begin with "they were genuinely brilliant"?

Because it determines what you learn from the story.

If you believe they were fools, the lesson is "don't be a fool." That lesson is useless, because nobody thinks they are being a fool. You will finish the book feeling safe, and then make exactly the same error.

If you accept that they were smarter than you — which is almost certainly true — then the lesson becomes: this failure had nothing to do with intelligence. It was structural. And structural things apply to you too.

Let us be blunt. The overwhelming majority of people reading this do not have Scholes's mathematics, Meriwether's trading instinct, or a Fed Vice Chairman's macro view.

If that group could lose $4.6 billion in four months, then "I will be more careful" offers you no protection whatsoever.

Division of labor with the rest of this library

This needs stating clearly, because the library already contains material on extreme risk.

Book	Owns
Antifragile & The Black Swan	The theory — why tail risk is systematically underestimated, what convexity is, what the barbell is
Way of the Turtle	The position math — expectancy, volatility-based sizing, pyramiding rules
This book	The forensic report — what actually happens when correlation really does go to 1 while you are levered. With the specific numbers and the specific dates

Taleb tells you why tail risk matters. The Turtles tell you how much to bet.

This book tells you what it looks like inside the account when the theory is right, the sizing model is sophisticated, and the world declines to cooperate anyway.

It does not re-derive convexity. It supplies the autopsy.

What the Fund Actually Did

LTCM's core strategy was convergence arbitrage, also called relative value arbitrage.

The logic is very clean, and worth understanding completely — because the logic is correct.

Imagine two bonds. One is a 30-year Treasury the government issued this week, called the "on-the-run" bond. The other is a 30-year Treasury issued six months ago, the "off-the-run" bond.

These two bonds carry identical credit risk — both are obligations of the U.S. government, with the same probability of default. Their maturities differ by six months. By any rational measure their yields should be nearly identical.

In practice they are not. The newer bond always yields slightly less — meaning it costs slightly more.

Why? Because the newer bond trades more actively and is easier to buy and sell. Institutions will pay a small premium for that liquidity. The gap might be a few hundredths of a percent.

LTCM's move was: buy the cheap off-the-run bond, short the expensive on-the-run bond.

Now notice what this position is. It does not bet on rates going up or down. If rates rise, both bonds fall; one side loses, the other gains, and they offset. It does not bet on the economy.

It bets on exactly one thing: that an irrational spread will converge.

And it almost always does converge. Because in six months, when the Treasury issues the next new bond, today's "on-the-run" becomes "off-the-run," its liquidity premium evaporates, and the spread closes by itself.

This is not speculation. It is harvesting a structural mispricing with a known expiry date.

So Where Is the Problem

The spread is a few hundredths of a percent.

That is the seed of the entire story.

Put $1 million into that trade and earn 0.05%, and you have made $500. After transaction costs, essentially nothing. That return cannot support a hedge fund, let alone two Nobel laureates' salaries.

There is only one way to turn a tiny but reliable profit into a large one: magnify it.

LTCM's leverage in normal times ran at roughly 25 to 1. Every dollar of its own capital controlled twenty-five dollars of positions. At certain moments, counting the notional exposure of its derivatives, effective leverage was far higher.

Now the trade becomes: $1 million of capital controls $25 million of positions, earns 0.05%, and produces $12,500. Against $1 million of capital that is 1.25%. Do it dozens of times and the annual return becomes extraordinary.

Which is precisely what LTCM delivered for three years:

Year	Return after fees
1995	roughly 43%
1996	roughly 41%
1997	roughly 17%

The first two years cleared 40% after famously steep fees. Wall Street's response was not skepticism but reverence. Banks competed to lend to them, on remarkably generous terms.

And those returns came almost entirely from leverage, not from genius at picking or timing.

An Arithmetic That Must Be Stated in Chapter One

The most misunderstood thing about leverage is the belief that it merely "magnifies gains and losses."

It does not merely magnify. It changes the size of the error you are able to survive.

At 25-to-1 leverage, your capital is 4% of your total position.

Which means: your positions need move against you by only 4% for your capital to reach zero.

Not halved. Zero.

Put that number on a realistic scale. A 4% adverse move is a small event in bond markets — one surprising inflation print can do it. And LTCM's spread positions were supposed to move far less than that. That was exactly why 25x leverage seemed defensible.

Their risk models showed that losing all their capital required an event that had never appeared in the historical record.

The models were not wrong. The historical data was not wrong.

In August 1998, the event that had never appeared, appeared.

Executable Trading Rules

This chapter gives you no strategy. It gives you three things you must accept before the later chapters can work.

1. Do not read this as "smart people do dumb things." That is the comforting misreading, and it lets you believe you are exempt. The correct reading is: this failure required nobody to do anything dumb.

2. Calculate your own zero point. Open your account and ask one question: how far do my positions have to move against me before my capital is gone? If you use no leverage, no margin, and sell no naked options, that number is 100% and you may rest. If that number is under 20%, you are on the same curve as LTCM — just at a different point on it.

3. Separate "the strategy is right" from "surviving until the strategy pays." LTCM's convergence trades were logically sound. Most of them did in fact converge, as we will examine in Chapter 5. The fund died anyway. Right but not solvent is, in outcome, identical to wrong.

Relevance to a Retirement Portfolio

This has to be said up front.

Nothing in this book is a strategy we suggest you run. Convergence arbitrage requires institutional funding lines, 25x leverage, and billions in scale. It is neither available to an individual investor nor worth pursuing.

This book's role in the library is inverted: it does not teach you a new method. It ensures that whenever you meet any leveraged product promising "low risk, steady returns," one number surfaces automatically.

$4.6 billion. Four months. Two Nobel laureates.

Fifty-seven places in this library discuss position sizing. This book covers what happens when position sizing rests on a hidden assumption, and that assumption fails.

What that assumption was is the subject of Chapter 2.

The core of your retirement portfolio should be low-cost index funds. Every chapter of this book makes that conclusion firmer.