The OG ‘Too Big to Fail’ - Long Term Capital Management and the Crisis of September 1998

Before Lehman. Before AIG and Bear Stearns. Before Fannie and Freddie. Before TARP and bailouts became part of the American political lexicon, there was a hedge fund in Greenwich, Connecticut, whose vulnerability to collapse was considered so dangerous that the Federal Reserve of New York commanded a meeting of the most powerful financial institutions in the world and told them — told them, they did not ask — to write checks and stated dollar amounts. Technically, no taxpayer money changed hands, but the template was established: private losses, socialized risk. The idea of "too big to fail" or maybe "too connected to fail" was established. At least the people responsible, the people who were supposed to be the smartest guys in the room ... any room ... lost everything in the process. It’s only fair that those who nearly burned down the house went up in smoke.

There is a seemingly particular troika of personality traits common in financial failures … Big Brain. Big Ego. Big Mouth. It may seem strange to imbue a hedge fund with a personality, but Long Term Capital Management (LTCM) of old money, white-shoe Greenwich, Connecticut, had one, and it embodied this lethal troika.

The Brains

LTCM was the brainchild of Wall Street legend, John Meriwether. In the 1980s, Meriwether built the bond arbitrage desk at Salomon Brothers into the most profitable operation on Wall Street. His traders generated extraordinary returns through a simple but powerful insight: mispriced securities tend to converge toward their theoretical value. So, buy the cheap ones. Short the expensive ones. Wait, collect, repeat. He was made famous in Michael Lewis’ first book - Liar’s Poker. The arbitrage team at “Solly” used to play a game of bluff and bravado called Liar's Poker. Players bet on the serial numbers on dollar bills. It’s a bit like trading. And nobody was better at it than Meriwether. It irked Solly CEO, John Gutfreund, that Meriwether (technically his underling) commanded such respect among the traders. So, one day, according to Lewis, Gutfreund approached the trading desk and birthed a Wall Street legend and created a visceral part of Salomon’s corporate identity.

He challenged Meriwether to Liars Poker. Mano a Mano: “One hand, one million dollars, no tears.”

Meriwether, without missing a beat, replied: “No, John, if we’re going to play for those kinds of numbers, I’d rather play for real money. Ten million dollars, no tears.”

Gutfreund guffawed, declared Merriweather crazy, and stomped off; apparently unaware that he’d just played … and lost.

Meriwether left Salomon Brothers under a cloud. He had, unwittingly, become embroiled in a Treasury auction scandal. One of his subordinates got caught placing falsified bids in government bond auctions, and when it came time for a corporate head to roll, Meriwether’s was it. He spent a couple of years regrouping, and then did what an indomitable spirit always does after a public humiliation. He went looking for redemption. He was still the smartest guy in the room and intended to prove it — and prove it big.

Long Term Capital Management launched in February 1994 with $1.25 billion in capital. It was the largest hedge fund launch in history at that point. The minimum investment was $10 million and investors had to commit to keeping their money in for at least three years. People lined up anyway. The names on the door were that powerful.

Myron Scholes and Robert Merton brought something no other fund could offer: genuine, peer-reviewed, Nobel-validated intellectual credibility. Their work on options pricing, the Black-Scholes-Merton model, had literally transformed global finance. Every derivatives desk on earth was using their framework. When they said the models worked, it carried weight that went beyond normal persuasion. It was like having the Pope as your parish priest. The Nobel committee awarded them the prize in October 1997 — while LTCM was running. The timing felt like divine confirmation.

David Mullins had been Vice Chairman of the Federal Reserve under Alan Greenspan. His presence at the LTCM table meant the fund had a former senior Fed official who knew every important person in global financial regulation personally. This was not incidental to how the fund raised money and conducted business. It was a feature.

Victor Haghani was the senior partner who ran much of LTCM's day-to-day trading. A graduate of the London School of Economics, he was Meriwether’s head trader back at Solly and a genuine quant. He was a pioneer of the bond-arbitrage strategy that made Meriwether an icon.

The Ego

LTCM's core approach was convergence trading, also called relative value arbitrage. The formal details of the strategy are a dense thicket of math. But the idea is straightforward… Take two securities that are theoretically identical (or nearly so) in terms of what they will ultimately pay. For various reasons, one trades at a premium to the other. The premium is temporary. Eventually the gap closes. Buy the cheap one, sell the expensive one, and collect the difference when convergence occurs.

The classic LTCM trade was “On-the-Run” versus “Off-the-Run” Treasury bonds — that is, the newest issued bonds (on-the-run) vs bonds of the same maturity but issued months ago (off-the-run). When the Treasury issues a new 30-year bond, it immediately trades at a slight premium to the previous 30-year bond — the one issued six months ago — because it is more liquid. Traders want the new one. But in thirty years, both bonds pay exactly the same thing. The price difference is pure liquidity premium, and it eventually evaporates. LTCM would buy the old bond, short the new one, and wait.

The problem with this trade — and with every convergence trade — is that the spreads involved are tiny: fractions of a percent. To generate returns that justified the fund's fees and the partners' ambitions, LTCM had to do these trades at enormous scale. And this meant leverage. Lots and lots and lots of leverage. At its peak in early 1998, LTCM had approximately $125 billion in assets on roughly $4.7 billion of equity capital. That's leverage of about 25 to 1. But that number dramatically understates the true exposure because it doesn't account for the derivatives positions (the swaps, the options, the forward contracts) that sat off the balance sheet. When those were included, LTCM's economic exposure was estimated at somewhere between $1 trillion and $1.25 trillion. To put that number in context, the entire US federal budget in 1998 was $1.65 trillion.

The numbers were eye-watering. But the models said this was fine. More than fine, actually. The models said the risk was precisely quantifiable and well within manageable bounds. Value at risk (VAR) calculations showed that a loss of more than $45 million in a single day was essentially impossible. The historical data supported this. The mathematics was rigorous. Two Nobel laureates had signed off on the framework. Meriwether, Scholes, Merton, Haghani... these were genuinely exceptional financial minds. The convergence trading insight was real. They’d already made billions at Solly doing the same things. The models were sophisticated. The intellectual foundation was sound within its assumptions. More than sound. They represented a kind of certainty. And when you are certain — not just confident, but mathematically certain, Nobel-Prize-certain — that your trades must converge, the rational response is to size positions as large as your financing will allow. The big ego converted a sound insight into a $1.25 trillion economic exposure. Uncertainty and fear create appropriate position sizing. Certainty blows position sizes out of all proportion. And the risk models said this was A-OK.

Unfortunately, the framework had two quietly catastrophic assumptions buried inside it: (1) That price movements follow a distribution similar to the normal bell curve. That extreme events … the fat tails … are vanishingly rare. And (2) That markets move in small continuous steps through all the intermediate prices. That when you need an exit, there will be one available. If price goes from 90 to 55, it passes through 89 and 88 and 87 on its way to 58 and 57 and 56.

Benoit Mandelbrot, the mathematician who gave us fractal geometry, had argued for decades that these assumptions were wrong. His original work on cotton prices in the 1960s had shown that financial markets exhibit what he called “wild randomness” rather than “mild randomness.” The tails are far fatter than the normal distribution predicts. Extreme events happen far more frequently than the models suggest they should. And critically, prices jump. They do not always pass through intermediate values. The exit priced at 88 is closed. Price sometimes goes directly from 90 to 55 without touching any price in between. There is a particular hubris in calculus training that Mandelbrot identified and that LTCM embodied perfectly. When your entire mathematical education is built around functions that are smooth and differentiable — where you can zoom in infinitely and the curve always looks well-behaved — you develop a deeply embedded intuition that the world works that way. The intermediate value theorem becomes an unconscious assumption about reality. Markets, it turns out, did not get the memo.

The Big Mouth

LTCM's early years were spectacular, with returns of 28% in 1994, 43% in 1995, 41% in 1996. Money was coming in over the gunwales. The fund was generating so much money so consistently that by 1997 Meriwether returned $2.7 billion to outside investors — not because performance was slipping but because the fund had gotten so large it struggled to find enough trades to deploy the capital. This should perhaps have been a warning sign. Instead it looked like discipline. And because it’s not very satisfying to be the smartest guys in the room if nobody knows, they boasted. They were justly proud of what they were doing and whom they were doing it with. So it was only a matter of time before everyone else on the street was copying their methods and trades. Spreads that were already notoriously tight narrowed still further as others crowded into the convergence trades. This necessitated ever larger positions to keep the profits growing.

The Thing That Can’t Happen, Happens

LTCM responded by expanding into new areas: equity volatility, merger arbitrage, emerging market debt. Areas where the partners were less experienced. Areas where the correlations they were assuming were less established and well understood.

July 1997: the Asian financial crisis begins. Thailand devalued the baht. Indonesia, Malaysia, South Korea followed. Spreads widened across emerging markets. LTCM took hits. The partners responded by adding to positions, convinced the models were right and the dislocations were temporary. This is the characteristic response of people who believe deeply in their own framework — when the market disagrees with the model, it’s the market that’s wrong.

May 1998: Salomon Brothers, Meriwether's old firm, announced it was shutting down its bond arbitrage desk. The desk that Meriwether had built and that had defined Wall Street fixed income trading for a decade. Salomon began unwinding its positions. Many of those positions were similar to LTCM's. Spreads weren’t converging. Spreads widened. LTCM lost $600 million in six weeks. Still not fatal. The partners believed the thesis was intact.

August 17, 1998: Russia defaults. The Russian default was not supposed to happen. Russia was a nuclear power, a permanent member of the UN Security Council. The IMF had just provided a substantial bailout package. The geopolitical establishment considered a Russian default essentially unthinkable. Too dangerous, too destabilizing, too embarrassing for the institutions that had been managing Russia's transition from Soviet communism. Russia defaulted anyway. And simultaneously devalued the ruble. In one announcement, it wiped out the value of the GKO bonds that investors around the world had bought specifically because they believed Russia was too important to be allowed to fail.

What happened next is the part that the models could not have anticipated. In a crisis of sufficient severity, correlations go to one. Everything that was supposed to be independent starts moving together, in the same direction, all at once. Every institution exposed to similar convergence trades — and many were, as LTCM's strategies were widely known and widely imitated across the industry — tried to unwind simultaneously. But there were no buyers. Spreads didn't converge. They exploded in the wrong direction. LTCM lost $550 million on a single day in August. The risk models confidently assured the partners that a loss of this size was impossible. But it happened anyway. By the end of August they had lost $1.9 billion, nearly half their equity capital. By late September they had lost $4.4 billion and were functionally insolvent on any realistic mark-to-market basis.

The trades, incidentally, were still on. The thesis had not changed. In fact, the spreads had widened so dramatically that the theoretical profit on convergence was now enormous. Still convinced, the partners put in their own funds to capitalize the firm. The problem was that LTCM had run out of time. They were right. They were just right at the wrong time— which, in finance, is the same as being wrong.

During the crisis, Warren Buffett, along with Goldman Sachs and the insurance giant AIG, made a formal offer to purchase LTCM's entire portfolio for $250 million and inject $3.75 billion of fresh capital. The offer gave the partners essentially nothing for their equity, but it resolved the crisis cleanly.

The partners rejected it. The terms were considered insulting given what the portfolio was theoretically worth if the trades converged as the models predicted.

Buffett's comment afterward became one of the most quoted assessments of the entire episode: "To make money they didn't have and didn't need, they risked what they did have and did need. And that's foolish. If you risk something that is important to you for something that is unimportant to you, it just does not make any sense."

LTCM's counterparties included every major bank and investment bank on Wall Street: Bear Stearns, Goldman Sachs, Merrill Lynch, Morgan Stanley, Lehman Brothers, JP Morgan, Bankers Trust, Société Générale, Deutsche Bank, UBS, Credit Suisse, Barclays, and more. Each of them was on the other side of various LTCM positions. Each had its own exposure that depended in various ways on how LTCM's positions were valued and ultimately resolved. If LTCM was forced into a disorderly liquidation, if $1.25 trillion of positions had to be suddenly unwound into markets that were already in crisis, the losses would not stay at LTCM. They would cascade. Every institution on the other side of those trades would take losses. Those losses would force further deleveraging and cause further price declines. And this would cause further losses. The doom loop that became familiar in 2008 was a genuine risk in September 1998.

William McDonough, President of the Federal Reserve Bank of New York, convened a meeting at the Fed's offices at 33 Liberty Street in lower Manhattan. He “invited” the chief executives of the fourteen major financial institutions most exposed to LTCM. The message was not subtle: you are going to recapitalize this fund because the alternative — a disorderly unwind — is worse for all of you than whatever you put in today.

Fourteen institutions wrote checks of roughly $300 million each, making total recapitalization approximately $3.6 billion. LTCM's partners were effectively wiped out, and the fund was now controlled by its creditors. The positions were unwound over the following months in an orderly way, and a systemic crisis was averted. No taxpayer money changed hands. The Fed did not write a check. But the Fed did convene the meeting, applied pressure, and made clear that the alternative was unacceptable. The line between public and private had been crossed in a new way, and the precedent was established: an institution could be so large, so interconnected, so embedded in the financial system that its failure could not be permitted regardless of the moral hazard implications.

Ten years later, that precedent would be industrialized on a scale that made LTCM look like a rounding error.

The Russian default was August 17th. The crisis reached its zenith (or nadir, depending on your perspective) in the third week of September 1998. The Fed bailout meeting was September 23rd. The fund passed into creditor control shortly after.

Twenty-eight years later, in a market carrying its own set of elevated valuations and risk overhangs — tariffs, geopolitical instability, a federal debt load that would have been unimaginable in 1998, inflation that has proven stickier than the models predicted — the LTCM story deserves to be retold.

Epilogue

Victor Haghani lost most of his personal fortune in the collapse. He spent the years afterward doing something unusual for someone in his position: thinking honestly about what went wrong rather than defending what had gone right.

The result is Elm Wealth and its public vehicle, the ELM Market Navigator ETF. Haghani is a co-founder. The philosophy is the explicit inversion of every LTCM mistake: no leverage, low cost, simple index instruments, systematic rules, complete transparency about the methodology, and explicit humility about the limits of any model's ability to forecast.

The fund's current portfolio holds approximately 25% in three-month Treasury bills. When the P-CAPE valuation model says equities look expensive relative to history, it moves money to cash and waits. It does not lever up to capture the theoretical profit on trades that must converge. It does not assume the exit will be available when needed.

The ETF’s large T-bill position is a direct descendant of September 1998. The man who lost $100 million because the models said the trades simply had to work built a fund whose first principle is that the models might be wrong and you need to survive being wrong.

That is an expensive education, but it produced a real-world lesson.

Note: I do not now, nor have I ever, owned ELM. But I do heartily endorse the philosophy and try to construct my own retirement portfolio accordingly. Priority number ONE is “Avoid Ruin.” When financial ruin is a legitimate possibility (and the LCTM story demonstrates it always is, no matter what the math says), do whatever is necessary to take that possibility off the table. Everything else—income, growth, taxes—is secondary.