To most people, the time it takes a honey bee to flap its wings is pretty much worthless. In the hands of Mani Mahjouri, however, those three milliseconds are a financial eternity.
Over the past decade, the U.S. government has imposed a series of rule changes on the structure of the country’s financial system. Markets that once moved at the speed of, well, humans now process transactions at speeds that are unfathomable to most of us. An entire, massive financial universe has formed in these flashes of time, just below the surface of everyday markets. It’s called high-frequency trading.
Mahjouri is the chief investment officer at Tradeworx, a high-frequency trading firm based out of Red Bank, N.J., that’s at the center of this revolution in trading. Tradeworx’s cutting-edge system for automatically executing high-frequency trades is responsible for one out of every 20 financial transactions on the planet.
“People still have this impression that, when they order a trade, the order goes down to the trading floor and a broker there executes it for them,” Mahjouri explains from his offices. “That’s the way stocks had been traded for 100 years. But that’s not the case anymore.”
For a significant portion of those trades, three milliseconds could be the difference between making a buck and watching it swirl down the drain. In the eyes of some of the practice’s critics, however, high-frequency trading has created a world where looking at a stock quote from home is like gazing at a star that burned out a million years ago.
The need for speed
One common misconception about the Internet is that physical geography has no effect on the way data moves. If the Web is an amorphous butt of ones and zeroes floating somewhere in the ether, why does it matter where a server is located?
But even on the Internet, distance is time. And time is money.
“What is happening with high-frequency trading is a new drain on social welfare.”
Imagine a soy bean trader on the floor of the Chicago Mercantile Exchange 50 years ago. If he could see a flurry of activity happening over in the Frozen Concentrated Orange Juice section, he might make an educated guess about his own stocks. His height could give him a momentary advantage over the shorter competitors standing right next to him. That’s why, as Mahjouri explains, financial firms used to recruit former college basketball players to work on the floor.
Nowadays, information may be traveling hundreds of miles instead of across a room. But it still has to physically get from point A to point B. No matter how efficient the connection, a shorter distance is always better.
That’s why, in 2010, a little-known fiber optics company Spread Networks spent $300 million constructing cables between Chicago and New York. The hope was that it could shave three milliseconds off the time it takes a piece of data to jump the 733 miles between the two cities. A similar effort to slice a whopping 60 milliseconds off the trip between London and Tokyo has a projected price tag in the neighborhood of $1.5 billion.
If you ask some high-frequency trading prognosticators, though, companies like Spread Networks are actually betting on the past. Fiber connections—glass tubes surrounded in plastic coating—have been an industry standard for years, but they might already be headed toward the technological scrap heap.
Smarter firms are banking on wireless microwave connections. The reason? Simple physics. Under ideal conditions, a ray of light in fiber goes 200,000 miles per second. A ray of light moving through the air goes 290,000 miles per second.
“With fiber, you never get anything optimum because there’s always going to be a slightly faster way to get there,” says Anova Technologies CEO Mike Persico, whose company is an industry leader in the construction of lightning-fast linkages for high-frequency trading firms. “But with wireless, if you have the right equipment and put it in the right locations, you can’t be beat.”
Companies that transmit over microwave must register their slice of the wireless spectrum with the Federal Communications Commission. Persico boasts that, in a few short years, Anova has gobbled up a larger portion of the high-frequency spectrum than any other single organization in the country—even more than brand-name telecom giants like Sprint or Verizon.
Others in his field are more cagey. Many register with the FCC under shell corporations with intentionally misleading names.
“You’ll see a company with a name like ‘Appalachian Gas Holdings’ buying up spectrum,” Persico says. “That’s clearly a shell company for someone that doesn’t want to be found out.”
According to Mahjouri, if one of Tradeworx’s rivals knew, down to the nanosecond, how fast the company could send a signal between New York and Chicago, it would then have an idea of how much faster it needed to be to beat Tradeworx. With that knowledge in hand, it would therefore know precisely how much money it would need to spend to do just that. By keeping everyone else pretty much in the dark, high-frequency trading firms have effectively turned much of the industry into a guessing game.
“I would love to know exactly how much a given boost in speed will increase our profitability. It would make my job a whole lot easier,” Mahjouri says. “But, the problem is, it’s not how fast you are. It’s how fast you are relative to everyone else.”
That battle for relative speed is sparking even crazier dreams straight out of science fiction: fleets of drones, lasers ricocheting through the stratosphere, and computer systems capable of making calculations faster than anything previously imagined.
A technological arms race
When Tradeworx was founded in late 1990s, the company wasn’t setting out to become a leader in the high-frequency trading space. In fact, in the beginning, it didn’t do any trading at all.
Instead, for its first few years Tradeworx simply provided informational tools, such as “limit order pricers,” to other investors—a strategy most famously used by Michael Bloomberg’s eponymous company to generate billions in annual sales.
After a few years, however, Tradeworx ran headfirst into the first of the two crises that would shape its history: the collapse of the dot-com bubble. Like so many other tech companies of the era, Tradeworx struggled to secure venture capital in a financial ecosystem where billions of Wall Street’s dollars had already been set ablaze.
“We decided, as a method of survival, to use our tools to make trades ourselves,” Mahjouri recalls.
In the early 2000s, Tradeworx began investing in the market directly and even went as far as creating its own hedge fund. The new strategy worked well—at least until Lehman Brothers, the investment bank where Tradeworx did much of its banking, became the first casualty of the 2008 financial crisis.
Overnight, the company lost its ability to make trades.
With the financial world around them once again falling apart, Tradeworx pivoted once more.
In the mid-2000s, the U.S. Securities and Exchange Commission had instituted a number of rule changes. The agency, for instance, began requiring stock prices on the 13 exchanges within the National Market System be listed electronically and for mandated quotes be priced down to the penny. This eliminated the old system where people would buy, for example, a share of AT&T for $35 ⅛. Now it could sell for $35.12 or $35.13. Another new rule required all stocks be sold for the lowest price offered.
These electronic listings allowed high-frequency traders to write algorithms that could gobble up vast quantities of data. Pricing stocks to the penny allowed fluctuations in a stock that would have previously been too small to register. In the past, for instance, something big would have needed to happen for AT&T’s stock to move from from $35 ⅛ up to $35 ¼, the smallest price differential allowed. Now even the financial equivalent of a slight breeze could push it from $35.12 to $35.13. And every time the price of a stock changes even the slightest, it’s an opportunity for a high-frequency trader to make money.
The problem is, it’s not how fast you are, it’s how fast you are relative to everyone else.
It was that final rule change—that all stocks go to the lowest bidder—that really kicked of the speed race. With so many companies feeding tons of info into their systems to find the lowest possible price to make a trade profitable, being first to a transaction was suddenly paramount.
In a 2013 paper, University of Chicago Booth School of Business economics professor Eric Burdish lays out a perfect example of how high-frequency traders utilize their speed advantages to make huge profits.
Burdish gave the example of two securities—let’s call them Security X and Security Y—that both track the S&P 500. Since they contain exactly the same basket of stocks, X and Y are going to rise and fall almost exactly in unison. When you zoom in to the scale of a few milliseconds, that “almost” becomes a massive gulf with one of the two securities jumping a penny or two for a fraction of a second before rejoining its near-identical twin.
“Market correlations that function properly at human-scale time horizons completely break down at high-frequency time horizons,” Brudish writes.
During the interval, an algorithmic trader armed with a high-speed connection can recognize the price differential, purchase the lagging stock at the cheaper price, hold it for a few milliseconds until the prices inevitably equalize, and then sell it for a small profit. The study estimates that the price differentials for these two alone occur thousands of times per day and are worth an average of $75 million per year. These sorts of money-making opportunities are constantly present in criss-crossing network of connections between the over 50 trading centers around the globe. And that makes the profit-generating opportunities provided by high-frequency trading almost limitless.
If this situation seems like a license to print money, it is—but only to a point. Between 2005 and 2011, the median duration of these gaps fell from 97 milliseconds down to seven, due to greater competition from an increasingly technologically savvy contingent of high-frequency traders. However, the profitability of the practice has remained remarkably constant. That means companies are having to buy bigger and bigger forks in order to gobble up the same size slice of pie.
In fact, the barges and drones and other sci-fi tech that high-frequency traders dream about are simply part of a debilitating trend in the industry: a technological arms race. While high-frequency trading has dropped from 61 percent of all trading volume in 2009 to 49 percent today, the dollar figure spent by the industry on technology has jumped from $400 million to $700 million. Already waves of firms have shut their doors or merged with rivals as they struggle to keep up.
“Like any arms race, the result is a cycle of spending which leaves everyone in the same relative position, only poorer,” writes Rick Bookstaber, a high-frequency trading skeptic. “Put another way… what is happening with high-frequency trading is a net drain on social welfare.”
Even Mahjouri, whose company not only conducts high-frequency trading but also sells access to its technology to other financial firms, isn’t comfortable with the implications of the seemingly never-ending quest for speed.
“We don’t like this game of spending all our time trying to be faster than everyone else. We’d rather spend our time trying to do things like determine if Microsoft is overpriced,” he says. “We’d rather work out better models then compete in the latency arms race.”
Small margins, high stakes
Sal Arnuk and Joseph Saluzzi, authors of the anti-high-frequency trading tome Broken Markets, argue that high-frequency trading “functions kind of as a tax on investors, who pay more for stocks they buy and receive less for stocks they sell.” According to Arnuk, high-frequency trading firms combine their lightning-fast reflexes with specialized data feeds purchased from the exchanges themselves. This, in turn, allows them to “backward engineer institutional investors’ footsteps and step in front of them all day long.”
In other words, these data feeds give high-frequency traders an early peek at orders coming down the pipeline. For smaller stocks, even relatively modest orders could shift the price up or down. An efficient high-frequency trader can hop in front of an investor’s order, gobble up the stock in question, then sell to the same investor for a small profit. Rinse and repeat hundreds of thousands of times per day, and you end up with a big win for high-frequency traders and a big loss for everyone else.
Looking at a stock quote from home is like gazing at a star that burned out a million years ago.
Arnuk argues this type of behavior doesn’t actually add any real value to markets. It doesn’t, for instance, help match a business that needs money with investors who have money to lend. Nor does it allow average people to save for retirement. Instead, it just lines the pockets of people sophisticated enough to take advantage of a broken system.
When it comes to high-volume stocks, such as Bank of America or Apple, Arnuk insists high-frequency traders are simply “playing a video game.” Exchanges offer small rebates any time someone offers a stock for lower than the current market price or offers to buy it for higher than the current market price. Conversely, markets charge fees for making trades outside of this range. This is supposed to make the markets move more smoothly because it decreases the “spread” between buyers and sellers.
These rebates, however, are naturally larger for companies that buy and sell a huge amount of stocks. That makes it effectively a transfer of wealth from average, everyday investors directly into the hands of high-frequency traders.
“This is a very profitable business,” Arnuk says.
While Arnuk is right that high-frequency trading has made a lot of people very, very rich, it’s hardly an infinite cash cow.
Large price fluctuations across markets tend to make high-frequency trading more profitable, explains Adam Sussman of the financial research firm TABB Group. However, as the economy has improved since the heady days of the financial crisis, markets have become comparatively less volatile. Over the past few years, the fluctuations haven’t been quite as large as they were in the immediate wake of the financial crisis.
The ever-progressing arms race is also making the industry prohibitively expensive. “Many of the smaller firms are disappearing and bigger ones are reducing staff,” Sussman says. “If your primary competitive edge is speed, you have less opportunity now than you did in 2009. Speed is always an important factor. But increasingly it’s not the make-or-break factor.”
This downscaling comes at a time when government regulators on both sides of the Atlantic are finally starting to seriously grapple with the issue. The U.S. Securities and Exchange Commission, long considered outgunned when it comes to policing high-frequency traders, acquired a computer system called Midas from Tradeworx that allows the agency to collect over a billion trade records per day and see the market in much the same way as high-frequency traders see it.
For its part, the European Parliament officially adopted a new set of rules last year that were labeled by European Commissioner for Internal Market and Services Michel Barnier as “one of the strictest set of regulations for high-frequency trading in the world.”
The rules, however, aren’t scheduled to go into effect 2016. They’re also not widely seen as doing anything that would put a definitive halt to the practice. Europe’s new regulations include requiring high-frequency trading firms keep records of their trading activities, prohibiting exchanges from discriminating on which firms they allow to colocate, and specifically identifying whose orders where placed algorithmically versus those made by flesh-and-blood traders. The EU’s regulations are largely based on those that went into effect in Germany in early 2014, and that law doesn’t seem to have put a major dent into high-frequency trading activities in the country.
Even still, for those who manage to stay in the game and adapt to the shifting regulatory environment, the large monetary rewards will still be there. Whether the path to all that money requires laser-guided drones or a technology only whispered about remains to be seen.
A version of this story was originally published on the Daily Dot on Nov. 15, 2013.
Illustration by Max Fleishman