Michael Crane had never met a problem that could not be solved by a sufficiently sophisticated model. He had a PhD in mathematics from MIT, a hedge fund called Crane Capital with four billion dollars under management, and a conviction, held with religious intensity, that the universe was ultimately quantifiable. In the spring of 1995, when he was thirty-nine years old and already rich enough to never work again, he was introduced to the only problem that would break him.

The Quant was a Belarusian named Viktor, thirty years old, with degrees from institutions no one had heard of and a reputation for building models that predicted market movements with an accuracy that was either genius or fraud. Michael did not care which. He paid Viktor a million dollars for a set of algorithms delivered on three encrypted floppy disks.

"The first disk," Viktor said, in English that was precise but unstressed in the wrong places, "will make you richer than your competitors. The second will make you richer than your investors. The third will show you how much all of it was worth."

The first disk predicted the collapse of a Japanese semiconductor manufacturer. Michael shorted it. The stock fell seventy percent in six weeks. Crane Capital made five hundred million dollars.

The partners celebrated. The investors celebrated. The press wrote profiles. Viktor's retainer was doubled.

The second disk predicted a currency crisis in Thailand, and then in Indonesia, and then in South Korea—a cascade of failures that would sweep through the Asian markets in 1997. Michael positioned his fund accordingly. When the crisis hit, Crane Capital made three billion dollars. Michael was on the cover of Fortune.

"What's on the third disk?" one of the partners asked at the celebration dinner.

"I haven't looked yet," Michael said. "It's for when I need it."

The third disk predicted that Crane Capital would be insolvent by the end of 1998.

Michael read the projection twenty times. The model showed a series of interconnected trades that would, through a chain of counterparty failures and margin calls, unravel the entire fund. The trigger event was a loan that Crane Capital had extended to a Russian bank, a loan that Michael had personally approved but that Viktor's model had been designed to avoid.

"I didn't make that loan," Michael told Viktor on the phone. "Your model didn't account for it."

"I know," Viktor said. "That is the point."

Michael began to short his own fund. He moved the trades through shell companies, through offshore accounts, through instruments so complex that even his own compliance department could not trace them. He was betting against himself, and every bet was paying off, and every payout was drawing him closer to the edge of illegality.

The partners discovered the shorts in August of 1998. They called an emergency meeting. They accused Michael of fraud, of self-dealing, of destroying the fund for his own enrichment. Michael tried to explain the model, the prediction, the Russian loan. They did not listen.

The SEC opened an investigation in September. The Russian bank defaulted in October. Crane Capital filed for bankruptcy in November.

At the hearing, Michael's lawyer asked Viktor to testify about the algorithms. But Viktor had disappeared. His retainer had been routed through a chain of shell companies that terminated in the Cayman Islands. His phone number was disconnected. The models on the floppy disks, when analyzed by forensic accountants, were found to contain nothing but random number generators linked to historical data.

"Mr. Crane," the SEC attorney asked, "did you base four billion dollars in investment decisions on a random number generator?"

Michael did not answer. He could not answer. He had followed the model because the model had been right, and the model had been right because he had followed it, and the distinction between cause and effect had dissolved somewhere in the billions of dollars that had passed through his hands.

He was sentenced to four years for securities fraud. He served two. When he was released, he went to work for a quantitative trading firm in Singapore. They paid him a fraction of what he had once earned. He did not complain. He understood, now, that the universe was not quantifiable. His mistake had been believing that it was. His punishment was knowing that it was not.

Viktor was never found. The algorithms were never explained. The only thing anyone could agree on was that a man who had once believed in numbers had been destroyed by them, and that there was, in the end, nothing random about it.

OTMES-v2-BLK-11-K3M7H8-E1240-M0-T014-9R5501-6D12