Man Vs. Machine: Inside the World of High-Frequency Trading
Kaufman’s interest in high frequency trading predates mine. When he was sworn into office in January 2009 to fill the Senate seat of his former boss Joe Biden, Kaufman was hell-bent on making sure that everybody responsible for the 2008 market meltdown paid for their actions. He soon focused on short-selling, urging the SEC to reinstate the uptick rule requiring short sales to be filled at a higher price; the rule had been eliminated in 2007.
He told me that when he was in business school in the 1960s, it was “an article of faith” that the uptick rule was “one of the two or three things that helped deal with predatory bear raids.” As a result of his interest in short-selling, Kaufman said, his office started getting calls from some fairly sophisticated people, including former Wall Streeters, telling him that if he thought that practice was bad, he should look at high frequency trading.
Kaufman likes to draw an analogy between high frequency trading and the swaps market. “With synthetic derivatives, you had a lot of money at stake, no transparency and then a major meltdown,” he explained to me. “If you look at high frequency trading, I think the same Kaufman formula works.”
A graduate of the Wharton School of the University of Pennsylvania, the 71-year-old Kaufman is a quick study and understands markets. If I were a high frequency trader, I’d take him seriously.
Kaufman has been high frequency trading’s loudest critic. But he’s far from alone. Seth Merrin, founder and CEO of Liquidnet Holdings, which operates an electronic marketplace that provides block trading for institutional investors, likes to compare high frequency traders to the American army during the Revolutionary War.
“The institutions are the equivalent of the British army, walking down the battlefield wearing bright red,” he told me back in March in his glass-enclosed office at Liquidnet’s sleek midtown Manhattan headquarters. “The high frequency traders are the Americans hiding in the woods in camouflage, picking them off. If the British army hadn’t changed its tactics, they would have lost every subsequent war.”
Even Duncan Niederauer, who as the pragmatic CEO of NYSE Euronext has been retooling his exchange to attract more business from high frequency traders, took a swipe at them. “We as an industry have to say how much is too much of this technology,” he said during an interview on CNBC after the flash crash, undoubtedly causing some consternation among the folks at NYSE Euronext who are selling space in the company’s new, 400,000-square-foot data center and co-location facility in Mahwah, New Jersey.
High frequency traders say that any efforts to rein in technology would be misplaced. Although speed is important to what they do, the quality of a firm’s computer models for analyzing markets and identifying where and at what price to buy and sell securities is what really determines success or failure, they argue. In their defense, high frequency traders say that they increase liquidity, lower trading costs, improve price discovery and reduce risk by dampening short-term volatility.
“High frequency trading is the liquidity backbone of the equity markets,” Manoj Narang, the founder, CEO and chief investment strategist of Tradeworx, told me when I first met him, in early March. “Long-term investors are the ones who cause bubbles, as well as liquidity crises when these bubbles burst.”
Narang, 40, is one of only a handful of proprietary traders I found willing to talk openly with a journalist about what they do. Most prefer to operate in the shadows, both to protect their valuable algorithms and to avoid regulatory scrutiny. But Narang, who left Wall Street in 1999 to start Tradeworx, sought me out when he heard through a public relations contact this winter that I was working on a story on high frequency trading.
His 25-person firm, which operates out of an office above a Restoration Hardware store in Red Bank, New Jersey, trades about 40 million shares a day on about $6 million in proprietary capital. Tradeworx also runs a $500 million statistical arbitrage hedge fund (which trades another 40 million shares a day) and owns a subsidiary, Thesys Technologies, which licenses its high-performance trading platform to other investors.
Narang lifted his profile on May 6 when he revealed to the Wall Street Journal that his firm turned off its high frequency trading computers during the flash crash. Tradeworx wasn’t the only one to do so. Kansas City–based Tradebot, started by BATS founder David Cummings, also stopped trading. Tradebot is one of the world’s two largest high frequency firms, reportedly trading as many as 1 billion shares a day in U.S. equities. Only Chicago-based Getco is thought to be bigger. Although Getco won’t comment on its daily trading volume, a spokeswoman for the firm did tell me that it continued to provide a two-sided market on all the electronic exchanges during the flash crash.
Most high frequency traders, in fact, kept their computers running, according to Jeffrey Wecker, president and CEO of Lime Brokerage. Wecker should know. His firm, which accounts for as much as 5 percent of the daily equity trading volume in the U.S., is the oldest and largest provider of high-speed trading solutions and access to all major U.S. exchanges for high frequency traders.
The high frequency firms that did stop trading on May 6 have been criticized for contributing to the decline by pulling liquidity from the market when it was needed most. But Narang told me that his firm had no choice because the exchanges were likely to cancel, or break, trades that were clearly erroneous (like selling Accenture at a penny a share). “If the exchanges broke all our buys and not our sells, we could have exceeded our capital requirements,” he explained. “We didn’t want to take the risk. The high frequency traders who continued to trade that afternoon made a fortune.”
In January, 2010, the SEC published a concept release on equity market structure, seeking public comment on everything from high frequency trading strategies and systemic risks to co-location and dark pools. At 74 pages, the report might seem like a real snoozer, but it’s actually a great primer on how the U.S. equity markets have responded to regulatory changes, starting in 1996 with the adoption of “order-handling” rules.
These new rules, which were designed to make the markets fairer following the Nasdaq price-fixing scandal in the mid-’90s, created ECNs and gave them the power to publish their stock quotes publicly alongside those of the listed markets. In 1999, Regulation Alternative Trading System (ATS) went into effect, enabling ECNs to operate as market centers without having to register as exchanges.
By the following year ECNs like Island and Archipelago had taken about one third of market makers’ volume in Nasdaq-listed stocks. But it wasn’t until after April 2001 — the deadline the SEC mandated for all U.S. exchanges to switch from fractions to decimals — that electronic trading really started to take off.
As bid-offer spreads shrunk and competition increased, ECNs and exchanges adopted a “maker-taker” pricing scheme to attract liquidity. Under the maker-taker model, market participants that offer to provide, or make, liquidity by posting an order to buy or sell a certain number of shares at a particular price receive a rebate. Those that execute against that order — that is, take the liquidity — have to pay a fee. Exchanges earn the difference between the rebate they pay and the fee they charge. The SEC limits taker fees to 0.30 cents a share; rebates tend to be lower for economic reasons, but for high frequency firms trading millions of shares a day, they can make for a pretty good living.