Flash Boys: A Wall Street Revolt (6 page)

The cost of RBC’s creating and running its own dark pool, Brad now learned, would be nearly $4 million a year. Thus his second question for the Golden Goose: How will we make more than $4 million from our own dark pool? The Golden Goose explained that they’d save all sorts of money in fees they paid to the public exchanges—by putting together buyers and sellers of the same stocks who came to RBC at the same time. If RBC had some investor who wanted to buy a million shares of Microsoft, and another who wanted to sell a million shares of Microsoft, they could simply pair them off in the dark pool rather than pay Nasdaq or the New York Stock Exchange to do it. In theory this made sense; in practice, not so much. “The problem,” said Brad, “was RBC was two percent of the market. I asked how often we were likely to have buyers and sellers to cross. No one had done the analysis.” The analysis, once finished, showed that RBC, if it opened a dark pool and routed all its clients’ orders into it first, would save about $200,000 a year in exchange fees. “So I said, ‘Okay, how else will we make money?’ ”

The answer that came back explained why no one had bothered to do any analysis on dark pools in the first place. There was a lot of free money to be made, the computer programmers explained, by selling access to the RBC dark pool to outside traders. “They said there were all these people who will
pay
to be in our dark pool,” recalled Brad. “And I said, ‘Who would pay to be in our dark pool?’ And they said, ‘High-frequency traders.’ ” Brad tried to think of good reasons why traders of any sort would pay RBC for access to RBC’s customers’ stock market orders, but he came up with none. “It just felt weird,” he said. “I had a feeling of why and the feeling didn’t feel good. So I said, ‘Okay, none of this sounds like a good idea. Kill the dark pool.’ ”

That just pissed off a lot of people and fueled suspicions that Brad Katsuyama was engaged in some activity other than the search for corporate profits. Now he was in charge of a business called electronic trading—with nothing to sell. What he had, instead, was a fast-growing pile of unanswered questions. Why, between the dark pools and the public exchanges, were there nearly sixty different places, most of them in New Jersey, where you could buy any listed stock? Why did the public exchanges fiddle with their own pricing so often—and why did you get paid by one exchange to do exactly the same thing for which another exchange might charge you? How did a firm he’d never heard of—Getco—trade 10 percent of the entire volume of the stock market? How had this guy in the middle of nowhere—in
retail
in
Canada
—learned of Getco’s existence before him? Why was the market displayed on Wall Street trading screens an illusion?

In May 2009, what appeared to be a scandal involving the public stock exchanges added more questions to Brad’s list. New York senator Charles Schumer wrote a letter to the SEC—then issued a press release telling the world what he had done—condemning the stock exchanges for allowing “sophisticated high-frequency traders to gain access to trading information before it is sent out widely to other traders. For a fee, the exchange will ‘flash’ information about buy and sell orders for just a few fractions of a second before the information is made publicly available.” That was the first time that Brad had heard the term “flash orders.” To the growing list of mental questions, he added another: Why would stock exchanges have allowed flash trading in the first place?

HE AND ROB
set out to build a team of people to investigate the U.S. stock market. “At first I was looking for guys who had worked in HFT or who had worked at large banks,” said Brad. No one who had worked in high-frequency trading would return his calls. Finding people who worked for the big banks was easier: Wall Street firms were shedding people. Guys who wouldn’t have given RBC a second thought were now turning up in his office begging for work. “I interviewed more than seventy-five people,” he said. “We didn’t hire any of them.” The problem with all of these people was that even when they said they had worked in electronic trading, they clearly didn’t understand how the electronics did the trading.

Instead of waiting for résumés to find him, Brad went looking for people who worked in or near the banks’ technology departments. In the end his new team consisted of a former Deutsche Bank software programmer named Billy Zhao, a former manager in Bank of America’s electronic trading division named John Schwall, and a twenty-two-year-old recent Stanford computer science graduate named Dan Aisen. He then set out with Rob for Princeton, New Jersey, where the Golden Goose resided, to figure out if any pieces of the Goose were worth keeping. There they found a Chinese programmer named Allen Zhang, who, it turned out, had written the computer code for the doomed dark pool. “I couldn’t tell who was good and who was not from just talking to them, but Rob could,” said Brad. “And it became clear that Allen was the Goose.” Or, at any rate, the only part of the Goose that might be turned to gold. Allen, Brad noticed, had no interest in conforming to the norms of corporate life. He preferred to work on his own, in the middle of the night, and refused to ever take off his baseball cap, which he wore pulled down low over his eyes, giving him the appearance of a getaway driver badly in need of sleep. Allen was also incomprehensible: What was just possibly English came tumbling out of him so quickly and indistinctly that his words tended to freeze the listener in his tracks. As Brad put it, “Whenever Allen said anything, I’d turn to Rob and say, ‘What the fuck did he just say?’ ”

Once he had a team in place, Brad persuaded his superiors at the Royal Bank of Canada to conduct what amounted to a series of science experiments in the U.S. stock markets. For the next several months he and his team would trade stocks not to make money but to test theories—to try to answer his original question: Why was there a difference between the stock market displayed on his trading screens and the actual market? Why, when he went to buy 20,000 shares of Intel offered on his trading screens, did the market only sell him 2,000? To search for an answer, RBC agreed to let his team lose up to $10,000 a day. Brad asked Rob to come up with some theories to spend the money on.

The obvious place to start was the public markets—the thirteen stock exchanges scattered in four different sites run by the New York Stock Exchange, Nasdaq, BATS, and Direct Edge. Rob invited the exchanges to send representatives to RBC to answer a few questions. “We were asking really basic questions: ‘How does your matching engine work?’ ” recalls Park. “ ‘How does it handle a lot of different orders at the same price?’ But they sent salespeople and they had no idea. When we kept pushing, they sent product managers, business people who knew a little about the technology—but they really didn’t know much. They finally sent developers.” They were the guys who actually programmed the machines. “The question we wanted to answer was, ‘What happens between the time you push the button to trade and the time your order gets to the exchange?’ ” says Park. “People think pushing a button is as simple as pushing a button. It’s not. All these things have to happen. There’s a ton of stuff happening. The data we got from them about what was happening at first just seemed random. But we knew the answer was out there. It was just a question of how to find it.”

Rob’s first theory was that the exchanges weren’t simply bundling all the orders at a given price but arranging them in some kind of sequence. You and I might both submit an order to buy 1,000 shares of Intel at $30 a share, but you might somehow obtain the right to cancel your order if my order is filled. “We started getting the idea that people were canceling orders,” says Park. “That they were just phantom orders.” Say the markets, together, showed 10,000 shares of Apple offered at $400 a share. Typically, that didn’t represent one person who wanted to sell 10,000 shares of Apple but rather a bunch of smaller sell orders lumped together. They suspected that the orders were lined up in such a way that some people at the back of the line had the ability to jump out of the queue the moment the people in the front of the line sold their shares. “We tried calling the exchanges and asking them if that’s what they did,” said Park. “But we didn’t even know what words to use.” The further problem was that the trading reports did not separate out the exchanges: If you tried to buy 10,000 shares of Apple that seemed to be on offer and succeeded in buying only 2,000 of them, you weren’t informed which exchanges the 8,000 missing shares had vanished from.

Allen wrote a new program that allowed Brad to send orders to a single exchange. Brad was fairly certain that this would prove that some, or maybe even all, of the exchanges were allowing these phantom orders. But no: When he sent an order to a single exchange, he was able to buy everything on offer. The market as it appeared on his screens was, once again, the market. “I thought, Crap, there goes that theory,” said Brad. “And that’s our only theory.”

It made no sense: Why would the market on the screens be real if you sent your order only to one exchange but prove illusory when you sent your order to all the exchanges at once? Lacking an actual theory, Brad’s team began to send orders into various combinations of exchanges. First NYSE and Nasdaq. Then NYSE and Nasdaq and BATS. Then NYSE, Nasdaq BX, Nasdaq, and BATS. And so on. What came back was a further mystery. As they increased the number of exchanges, the percentage of the order that was filled decreased; the more places they tried to buy stock from, the less stock they actually bought. “There was one exception,” said Brad. “No matter how many exchanges we sent an order to, we always got one hundred percent of what was offered on BATS.” Rob Park studied this and said, “I had no idea why this would be. I just thought, BATS is a great exchange!”

One morning, while taking a shower, Rob had another theory. He was picturing a bar chart Allen had created. It showed the time it took orders to travel from Brad’s trading desk in the World Financial Center to the various exchanges. (To widespread relief, they’d left Carlin’s old offices and moved back downtown.) “I was just visualizing that chart,” he said. “It just occurred to me that the bars are different heights. What if they were the same height? That got me fired up immediately. I went to work and went right to Brad’s office and said, ‘I think it’s because we’re not arriving at the same time.’ ”

The increments of time involved were absurdly small: In theory, the shortest travel time, from Brad’s desk to the BATS exchange in Weehawken, was about 2 milliseconds, and the slowest, from Brad’s desk to Carteret, was around 4 milliseconds. In practice, the times could vary much more than that, depending on network traffic, static, and glitches in the pieces of equipment between any two points. It took 100 milliseconds to blink your eyes; it was hard to believe that a fraction of the blink of an eye could have such vast market consequences. Allen wrote a program—this one took him a couple of days—that built delays into the orders Brad sent to exchanges that were faster to get to, so that they arrived at exactly the same time as they did at the exchanges that were slower to get to. “It was counterintuitive,” says Park. “Because everyone was telling us it was all about faster. We had to go faster. And we were slowing it down.” One morning they sat down at the screen to test the program. Ordinarily, when you hit the button to buy and failed to get the stock, the screens lit up red; when you got only some of the stock you were after, the screens lit up brown; and when you got everything you asked for, the screens lit up green. Allen hadn’t taken his Series 7 exam, which meant he wasn’t allowed to press the Enter button and make a trade, so Rob actually hit the button. Allen watched the screens light up green, and, as he later said, “I had the thought: This is too easy.” Rob did not agree. “As soon as I pushed the button, I ran to Brad’s desk,” recalled Rob. “ ‘It worked! It fucking worked.’ I remember there was a pause and then Brad said, ‘Now what do we do?’ ”

That question implied an understanding: Someone out there was using the fact that stock market orders arrived at different times at different exchanges to front-run orders from one market to another. Knowing that, what do you do next? That question suggested another: Do you use this knowledge to join whatever game is being played in the stock market? Or for some other purpose? It took Brad roughly six seconds to answer the question. “Brad said, ‘We have to go on an educational campaign,’ ” recalls Park. “It would have been very easy to make money off this. He just chose not to.”

THEY NOW HAD
an answer to one of their questions—which, as always, raised another question. “It’s 2009,” said Brad. “This had been happening to me for almost three years. There’s no way I’m the first guy to have figured this out. So what happened to everyone else?” They also had a tool they could sell to investors: the program Allen had written to build delays into the stock exchange orders. Before they did that, they wanted to test it on RBC’s own traders. “I remember being at my desk,” said Park, “and you hear people going, ‘OOOOOOO!’ and ‘Holy shit, you can buy stock!’ ” The tool enabled the traders to do the job they were meant to do: take risk on behalf of the big investors who wanted to trade big chunks of stock. They could once again trust the market on their screens. The tool needed a name. Brad and his team stewed over this until one day a trader stood up at his desk and hollered, “Dude, you should just call it Thor! The hammer!” Someone was assigned to figure out what Thor might be an acronym for, and they found some words that worked, but no one remembered them. The tool was always just Thor. “I knew we were onto something when Thor became a verb,” said Brad. “When I heard guys shouting, ‘Thor it!’ ”

The other way he knew they were on to something was from conversations he had with a few of the world’s biggest money managers. The first visit Brad and Rob Park made was to Mike Gitlin, who oversaw $700 billion in U.S. stock market investments for T. Rowe Price. The story they told didn’t come to Gitlin as a complete shock. “You could see that something had just changed,” said Gitlin. “You could see that when you were trading a stock, the market knew what you were going to do, and it was going to move against you.” But what Brad described was a far more detailed picture of the market than Gitlin had ever considered—and, in that market, all the incentives were screwed up. The Wall Street brokerage firm deciding where to send T. Rowe Price’s buy and sell orders had a great deal of power over how and where those orders got submitted. The firms were now paid for sending orders to some exchanges and billed for sending orders to others. Did the broker resist these incentives when they didn’t align with the interests of the investors he was meant to represent? No one could say. Another wacky incentive was called “payment for order flow.” As of 2010, every American stockbroker and all the online brokers effectively auctioned their customers’ stock market orders. The online broker TD Ameritrade, for example, was paid hundreds of millions of dollars each year to send their orders to a high-frequency trading firm called Citadel, which executed the orders on their behalf. Why was Citadel willing to pay so much to see the flow? No one could say with certainty.