The Kraken Project

12

Winter and summer, spring and fall, G. Parker Lansing had a fire going in the wood-burning fireplace in his office on the seventieth floor of One Exchange Place, in the Financial District of lower Manhattan. This little touch had cost him over $2 million to install. Then there were the seasoned, split birch logs that had to be brought up every day, on the service elevator, and carefully stacked in the nineteenth-century wrought-iron bin. Any one of his colleagues could stick a Cézanne on the wall. But a real wood-burning fireplace on the seventieth floor of a Manhattan skyscraper? That said something about him no painting could.

On this warm October day a fire did indeed burn merrily on the grate, the heat of it sucked away in a blast of air-conditioning. He was perched across from the fireplace, sitting at a Renaissance refectory table, at the focal point of a group of flat-panel computer screens arranged in a semicircle. One smooth, hairless finger poked out instructions on a computer keyboard. Lansing was a two-finger typist — he’d flunked the typing class in ninth grade. But he didn’t mind: typing was for secretaries and the working classes. Although G. Parker Lansing had been born in New York City, he thought of himself as a man more in tune with the culture and breeding of a member of the British upper class. He had even cultivated an accent to match.

Lansing was president and CEO of Lansing Partners, a third-tier boutique Wall Street firm. Lansing Partners specialized in the science of algorithmic or high-frequency trading. “Algo” trading, as it was often called, had taken over a major percentage of trading on the stock and commodities exchanges. In the year 2013, for example, 70 percent of all trades on the New York Stock Exchange were algo trades, automatically executed by computers with no intervening human decision making. These trades were made by computer algorithms, which received information electronically and then traded on that information within milliseconds, far faster than any human being.

Algo trading had been getting bad press for years. But as with anything on Wall Street, whatever made money would be allowed to continue until things blew up. People claimed it gave certain traders an unfair advantage. The financial markets, they said, should be fair. Lansing had nothing but contempt for those people who thought the markets were or should be fair. They deserved to lose their money. All the giant international banks and brokerage houses engaged in algo trading, at great profit, at the expense of the little people. And when the little people lost, it was their fault for being naïve. And when algo trading blew up the market — which it surely would some day — the government would bail them out. Private profit, public loss was the name of the game.

There was, for example, the algo trading program devised by Citigroup, called Dagger, which noted price differences between shares of the same company selling in, say, Hong Kong and New York; Dagger would then buy millions on one exchange and dump millions on the other, extracting a profit from the temporary spread — which might have lasted less than a second. Another famous algo program, called Stealth, devised by Deutsche Bank, parsed trades on the Chicago Mercantile Exchange, looking for statistical blips in the trading of oil futures; Stealth would go both long and short in the market and profit handsomely regardless of whether the market went up or down.

Who lost? The slow, dumb human traders. Ordinary investors. Retirement accounts. Pension funds. Towns and cities across America that had invested their paltry funds. Let us tip our hats, thought Lansing, to all the suckers, dupes, and mugs who thought the stock and commodities markets were playing on a level field.

The computers involved in algo trading had to be super fast, and they had to be physically close to the trading floor. Even the delay of a speed-of-light trade from, say, across the river in New Jersey might mean the difference between profit and loss. As a result, algo trading was used mostly by firms with offices located right in the Financial District, connected to the exchanges with fat fiber optic bundles going directly from their computers to the exchange computers.

G. Parker Lansing was Upper East Side born and bred — St. Paul’s, Harvard, Harvard MBA. He started out in the trading department at Goldman Sachs, where he designed algo-trading strategies. He didn’t actually write code — he knew little about the inner workings of computers. The code was for others to devise. His role was to identify trading opportunities and work out the strategies. He had designed dozens of algo-trading attacks for Goldman — programs that prowled the markets, sniffing out anomalies, seeking inefficiencies in buy-sell spreads, looking for stupidities, and identifying microdislocations in the price of everything from pork bellies to gold. The money to be made was stupendous. At Goldman Sachs he was beautifully compensated. He went the usual route of those in his class: the Trump Tower penthouse, the twenty-five-thousand-square-foot “cottage” in the Hamptons, the Greenwich mansion full of Damien Hirst dot paintings. And, of course, the Cayman Islands bank accounts and shell companies that brought his income tax rate well below that of the poor schmuck who mowed his four-acre lawn in East Hampton.

About four years ago, Lansing had had an algo-trading idea that was so brilliant and original that he’d decided not to share it with Goldman Sachs. Instead, he’d gracefully left that firm’s employ and started Lansing Partners. Once he had mapped out the strategy of this algo-trading idea, he looked around for a programmer. That was when he found Eric Moro, one of the founders of the shadowy hacker collective known as Johndoe. Moro was a man with a perfect combination of genius and flexible ethics.

At St. Paul’s, Parker Lansing had been a loud, braying bully with slicked-back hair. He and a group of friends spent much of their time pushing around fags, pussies, and tards. In college, he began to realize that his arrogant, frat-boy persona, which had worked so well for him in prep school, would be a disaster in real life. It would never get him where he wanted to go. And so, with great effort and perseverance, he had remade himself into a cultured, well-bred, well-dressed, deliberative young man with a faint British accent. Most important, he had also come to realize that, contrary to what his parents had taught him, Upper East Side WASPs weren’t the only worthy and intelligent people in the world. The really smart ones, in fact, tended to be ethnics — Jews, Polacks, Indians, Italians, Irish, Chinese. Moro was one of these smart ethnics, a greasy-haired Italian kid from some nondescript town in New Jersey, born into a hideous working-class family of cops and firemen. He had a Tony Soprano accent but no cool Mafia connections. Moro had somehow come out of that background a useful genius. Lansing did not use the word lightly: Moro was a true genius, and Lansing paid him accordingly.

Lansing’s brilliant idea involved a special kind of algo trading. To implement it, Lansing had designed, and Moro had coded, a unique algo program he’d nicknamed Black Mamba. The black mamba was the world’s deadliest snake, one of the few animals that would actually hunt down and kill a human being. It could slither faster than a running man and strike three times a second. Each bite would inject enough venom to kill twenty-five people. The Black Mamba program was, like its namesake, a fearsome and deadly hunter. It did one thing only: it was a bot that stalked and preyed on other algo programs. It lurked in the black pools of the markets, parsing millions of trades, until it found a mark — another algo program at work. It watched the program in action, figuring out its trading strategy. And when Black Mamba found an algo program engaging in a predictable trading strategy, it moved in for the kill. Knowing what the algo program was buying or selling, it would anticipate the mark’s trades, preempt them, and profit. Large mutual funds often used algo programs to break down one huge trade into hundreds of smaller trades, executed over a period of hours, sometimes on different exchanges. The goal was to keep the big trade a secret so as not to drive the price of the stock up or down. Knowing what the mark was going to do beforehand, Mamba would buy those same little blocks of stock a thousandth of a second before the mark put in its order. It would then sell those shares to the mark a split second later, at a profit. It might do this thousands of times before the owners of the algo program realized something was awry. But by then, of course, it was too late.

Over the past years, Mamba had grossed $800 million for Lansing Partners. But it was not an invulnerable strategy. Some of the large banks and hedge funds had become aware of Mamba’s activities and did not appreciate being conned at their own dirty game. They had tried to counter it. But Eric Moro tweaked Mamba almost daily. When Mamba made a mistake and lost money, Moro fixed it. When it used a strategy that no longer worked, Moro devised a new strategy. Like an ever-evolving virus, Black Mamba changed its attacks and even its basic coding, so that it was never recognizable from one week to the next.

On this particular Monday morning in mid-October, G. Parker Lansing was monitoring the progress of Mamba as it cruised a “black pool” trading exchange. That very morning, Mamba had identified a sucker of an algo program selling insider shares in a celebrated dot-com company. Exactly ninety days before, this dot-com company had had an IPO, an initial public offering of stock. Ninety days after an IPO, the firm’s insiders became eligible to sell their shares. It was an old story, from Facebook to Groupon. The insiders — the company founders and venture capitalists — cashed out after ninety days, leaving the suckers who’d bought the IPO with devalued stock. Today marked the ninety-day expiration. Today, it looked like the insiders would be selling big-time, but quietly, on the sly, using an algo program.

It appeared that a massive number of shares were going to be dumped that day in blocks of two to five thousand by company insiders. The dumb little algo program the insiders were using was disguising the trades, making them look like trades from many different individual investors. But the sudden activity, exactly ninety days after the IPO, was a tip-off. On top of that, the sloppiness of the algo program was causing the stock to trend down. Other traders were beginning to notice. As a result, the algo program was increasing its selling, trying to dump as many shares as possible before the price got even lower.

G. Parker Lansing felt his salivary juices flowing. This was as plump a goose as you could ask for, just ready to be slaughtered, plucked, and roasted.

Lansing unleashed Mamba and then sat back to watch the action. Mamba’s strategy this time was known as a “naked short.” The program would start selling shares of the dot-com stock that it didn’t own. This was not illegal. If the stock price went down — which it surely would — then Mamba would buy from the algo program the same number of shares Mamba had earlier sold but didn’t own. The beauty of the strategy was this: by selling shares it didn’t own, and then buying the same number of shares a few minutes later at a lower price, Mamba would square the balance sheet without spending a dime — and pocket the difference. The shares would be delivered to the buyer as part of a normal trade settlement pattern. In this way, Mamba would have disguised the fact that it had sold something it didn’t own.

Again, as with anything that made money on Wall Street, no matter how sketchy, this was All Perfectly Legal. It was called naked short selling, and it was extremely profitable — as long as the share price continued to go down.

As Lansing watched, Mamba struck. Of course, Lansing would not see the actual trading in real time, as it was taking place in a black pool at high speed — but Mamba would report to him the results as the trades were settled.

In a matter of seconds, Mamba sold sixteen million shares of the dot-com stock on the open market. These were shares it did not own: a naked short. Then it waited for the algo program to respond by offering block after block of stock for sale as the price sank, which Mamba would buy up at ever lower prices — locking in bigger and bigger profits.

Lansing stared at his screen, waiting for the dot-com stock to tick down, if not collapse. But nothing of the sort happened. Instead, the stock suddenly began to climb in value. And climb.

Lansing could hardly believe it. It made absolutely no sense. Suddenly, without warning, the insiders desperate to dump the stock had suddenly reversed and started buying — at higher and higher prices! Why? In desperation, Lansing tried to shut down Mamba. But it was too late: Mamba had already taken a naked short position on sixteen million shares. It couldn’t be undone. And the “dumb” algo program, instead of continuing to sell shares in a predictable fashion, had done something totally crazy. It had suddenly reversed itself and purchased Mamba’s entire naked short, on margin, driving up the price. And then it did something even crazier: it dropped out of the black pool and started purchasing huge blocks of the stock openly, on NASDAQ, for the whole world to see, driving the price up even more.

As a result of all these unexpected machinations, the stock shot up 30 percent in a matter of seconds. Which meant that to deliver the sixteen million shares of stock that Mamba had sold (but didn’t own), Lansing Partners would have to purchase those shares at a price 30 percent higher than what it had sold them for.

It was a classic short squeeze, the most painful and dreaded thing that could ever happen to a trader. Thus it was that ninety seconds into the trade, G. Parker Lansing was staring at a $320 million trading loss in the stock, a loss that was still rapidly increasing as the stock price went up—and there was nothing he could do about it. He had to “cover” his position by buying the sixteen million shares he had already sold but didn’t actually own. And as he moved to make the purchase before the stock got even more expensive, his own forced “cover” caused the shares to rise another 15 percent, killing him even more.

There was no way out of a classic short squeeze, no way to unwind the trade, no way to escape the loss. Lansing was completely crushed by the market. He stared as Mamba was forced to purchase the final block of stock at a 46 percent gain.

In 120 seconds, it was all over. G. Parker Lansing had lost $411 million.

He collapsed back into his chair. His hands were shaking. His mouth was dry. He could hear the blood squeaking in his ears. How had it happened? How had that dumb-ass algo program reversed itself and done such a crazy, illogical, unexpected, and bizarre trade? It was brilliant beyond belief, but such a crazy trade would only have been possible if the algo program knew exactly what Black Mamba intended to do beforehand.

Once he had framed the question that way, the answer became obvious. This was no accident. This was no black swan event. Black Mamba had been targeted. The “dumb” algo program had been written specifically to lure Mamba into a massive, risky trade and then spring a trap and short squeeze it to death.

Even as these thoughts coursed through Lansing’s mind, he heard an excited voice in the hallway. Eric Moro, his long-haired, jean-clad young partner, came bursting in, a look on his hollow face of total freak-out.

“What the hell? What the hell?”

Lansing held out a spidery hand. “Have a seat, Eric.”

“Aren’t you looking at your screens? Did you see what just happened?”

Lansing continued to hold out his hand. “A seat, please.”

“I wanna know what just happened here!”

“It’s quite simple,” said Lansing quietly. “We’ve been the victim of a sting operation.”

Moro stared, comprehension blooming on his face.

“You really need to sit down.”

Moro lowered himself into a massive leather armchair, releasing a gush of air.

Lansing went on, his voice calm and soothing: “Our task now is to find out who did this to us and take appropriate action.”

“Appropriate action? Like what? What’s ‘appropriate action’ against some turd-sucking dirtbag who just jacked us for four hundred million dollars?”

“Something so terrible, so awful, that no one will ever think of doing anything like this to us again. Only then”—Lansing smiled coldly—“will our business be truly secure.”