History never repeats itself in financial markets, but it often rhymes.
Long-Term Capital Management is a good example. Founded in 1994 by some of Wall Street’s brightest minds and built on the celebrated Black-Scholes-Merton model, the hedge fund briefly appeared untouchable. Then came Russia’s financial crisis in 1998. LTCM came to the brink of collapse, prompting the Federal Reserve Bank of New York to coordinate a 3.6 billion dollars rescue by fourteen major banks. One institution refused to participate: Bear Stearns.
The irony came later. In 2007, Bear Stearns itself was running two highly leveraged funds that had generated impressive returns for years with the help of sophisticated risk models. When the U.S. housing market turned, both funds unraveled within weeks. A year later, Bear Stearns became one of the first major casualties of the global financial crisis.
What destroyed LTCM was not the brilliance of its model. It was the leverage underneath it.
The model that underestimated the tail
That same lesson resurfaced with Lehman Brothers. Once again, the Value at Risk (VaR) model took center stage. Designed to measure risk, it assumed a well-behaved statistical distribution and underestimated the probability of extreme market moves—the so-called fat tails.
During the 2008 credit crisis, Wall Street mathematicians discussed this phenomenon. They described events that, statistically speaking, should occur only once every ten thousand years. Yet in just one week, they experienced three of them. Some even spoke of a pattern in their charts that, according to theory, could not exist—the so-called convexity smile.
The model was mathematically sound. Reality simply refused to cooperate.
And once again, leverage turned a bad outcome into a disastrous one.
AI’s turn as the holy grail
Today, the latest genius has a new name. In 2024, a former OpenAI researcher published a widely discussed essay about the rise of intelligent machines. His vision made such an impression that some began calling him the Nostradamus of AI. He later launched an investment fund built entirely around that AI thesis. This time, artificial intelligence had become the holy grail for sophisticated investors.
Then, a few days ago, it unraveled. Of what had been roughly $45 billion, only a fraction remained within days. The reported cause was leverage of as much as 400 percent.
When AI stocks declined, margin calls followed—the demand to post additional collateral immediately. He was forced to liquidate positions at steep discounts.
He may still prove completely right about AI over the long run. But in financial markets, you first have to survive the short term. The fund was called Situational Awareness. Ironically, the only situation that truly mattered escaped its founder.
Untouchable—until leverage strikes
What these stories have in common is more than just a clever model. First, look at the aura surrounding them. LTCM was led by Nobel Prize winners. Just before its collapse, Bear Stearns was still hailed as Wall Street’s most admired securities firm, in a ranking that specifically assessed the quality of its risk management. And Aschenbrenner was regarded as the Nostradamus of our time. In each case, the central figure seemed untouchable. And in each case, it was precisely that apparent invincibility—combined with leverage—that was invariably punished.
That is the most dangerous feature of leverage: you surrender your own autonomy.
You no longer decide when to sell. Your lender does. Usually at the worst possible moment.
When that happens, the casino rule applies without mercy: rien ne va plus. The money is no longer yours. With leverage, you actively seek out fat tails and skewed probability distributions. Financial markets seem almost designed to humble precisely those investors who are most convinced of their own certainty.
The tail wags the dog
Under normal circumstances, your investment returns are determined by your own decisions—your analysis, your patience and your conviction. The dog wags the tail. But with substantial leverage, market sentiment determines your outcome. A temporary decline can become fatal, even if your long-term investment thesis is correct. The tail wags the dog.
You can be entirely right and still lose everything because you no longer control when the game ends. Without leverage, you remain in control and can stay invested long enough to be proven right.
There is good news as well. Once forced selling has run its course, what remains is a cleaner market. That is often when the recovery begins. The fourteen banks that took over LTCM’s positions in 1998 earned roughly 10 percent over the following year because they bought when everyone else was forced to sell. Likewise, the Great Financial Crisis of 2008 became the starting point for one of the strongest bull markets in history.
Now that leverage is being forced out of the AI trade, the long-term outlook for AI stocks may once again be bright.
For investors who did not borrow money, today’s fire sale may well become tomorrow’s opportunity.
Han Dieperink is chief investment officer at Auréus Asset Management. Earlier in his career, he served as chief investment officer at Rabobank and Schretlen & Co.
