"Look," says Sergei Tchetvertnykh, pointing at a flashing spreadsheet on his desktop's screen. "I just made $82,000 in one second."
The co-CEO of the Toronto-based electronic trading firm Infinium Group isn't exaggerating. A second is now a very long time in financial markets, thanks to computer algorithms. Traders can gather and interpret market data, and buy or sell securities in response, in milliseconds (thousandths of a second) or even microseconds (millionths of a second).
Not every second is that successful, of course, and there can be many reversals over a few hours. On a typical day, Infinium, with offices in Toronto, San Francisco, London and Barbados, executes between 500,000 and one million trades of stocks, options, currencies and other financial instruments worldwide. Measured by volume of shares, it is often the largest single trader of major companies listed on the Toronto Stock Exchange-more active, in other words, than any of the otherwise dominant investment dealers owned by Canada's Big Five banks.
The paradox is that Infinium is still very small and very young, with 70-odd employees spread over its second-floor headquarters in a block of 19th-century buildings near Toronto's historic St. Lawrence Market. (Its other three offices account for another 40 staff.) Tchetvertnykh and co-CEO Alan Grujic, who are both 42, founded the firm in 2002. Unlike old-style investment dealers, Infinium is a pure proprietary trading outfit-it trades only its own money, none for clients.
As a specialist in high-frequency trading, Infinium is one of a handful of cutting-edge firms in Canada, alongside dozens more based in the United States and Europe, that have overwhelmed and revolutionized financial markets over the past few years. By some industry estimates, these hotshot dealers-along with Goldman Sachs and some other established firms that have also jumped into the high-frequency game-now account for about a quarter of daily stock trading volume in Canada, and as much as 60% to 70% south of the border.
Many of the strategies used by the high-frequency traders are traditional, such as arbitrage, which takes advantage of price anomalies in different markets. If, say, Barrick Gold is trading at $40.04 a share in Toronto and $40.05 in New York (a huge price gap these days), the high-frequency firm quickly buys in Toronto and sells in New York before the gap closes.
The high-frequency traders' speed and volume is scaring the daylights out of many regulators and traditional investment dealers, who think this new wave threatens to swamp the very foundations of financial capitalism. In November, Paul Myners, financial services secretary to the U.K. Treasury Department, told an interviewer that "the danger is that nobody really seems to think of themselves as owners." How can management be accountable to investors who change every few seconds? Thomas Caldwell, CEO of Caldwell Securities Ltd., a mid-sized Toronto dealer that has large investments in the NYSE Euro- next and other stock-exchange holding companies around the world, worries that "a lot of trading these days is disconnected from any economic reality or the fundamentals of companies."
High-frequency traders say that, far from bringing on the apocalypse, what they are doing is very safe and useful. If they buy and sell almost instantaneously at virtually the same price, the risk of massive losses is tiny. Moreover, they argue that huge benefits accrue to average investors in particular. "There's more liquidity and tighter spreads," says Grujic. "How can that not be better?" More liquidity means anyone can get an order filled almost immediately at the market price. (The "spread" is the formerly wide gap between the high price traditional brokers would quote to clients who wanted to buy a security, and the lower price they would offer to investors who wanted to sell.)
People forget, argue Tchetvertnykh and Grujic, just how clubby and antiquated stock and bond markets were as recently as the early 1990s, when the two of them entered the business. In those days, the Toronto Stock Exchange still had a trading floor, and the New York Stock Exchange (NYSE) accounted for more than 80% of all trading in the United States.
Of course, even Tchetvertnykh and Grujic had little idea of what the future of the markets would be when they got acquainted in 1992. The meeting place was the CAMI automotive factory in Ingersoll, Ontario. Grujic, who had graduated from the University of Toronto in 1990 with a bachelor's degree in electrical engineering, was programming and monitoring robots on the plant's assembly line. However, he'd decided to go to the University of British Columbia for an MBA, and was helping management look for his successor. One candidate was Tchetvertnykh, who had graduated from the Kiev Polytechnic Institute with a degree in cybernetics in 1990, and had come to Canada from Ukraine to enroll in the MBA program at the University of Western Ontario's Richard Ivey School of Business.
The two hit it off right away. Both were academically brilliant sons of European professionals. Tchetvertnykh's father was a physicist and his mother an accountant. Grujic was born in Toronto, but his parents were from the former Yugoslavia-his father an electrical engineer who founded his own consulting firm, and his mother a PhD in psychology. But the duo also realized that finance, not the professions, was the place to make big money in North America.
Tchetvertnykh graduated from Western's Ivey School in 1994, and was offered a job in corporate finance with Credit Suisse First Boston in New York. Grujic also graduated in 1994, but UBC wasn't on Wall Street recruiters' radar screen in those days, so he opted for a job in bond trading with TD Securities.
They could hardly have picked a more propitious time to enter the securities business. Markets around the world were on a roll and the tech boom was in full swing; investment banks were experimenting with new mathematically based trading strategies and starting to deal in more complex options and derivatives.
Tchetvertnykh specialized in international mergers and acquisitions, which inevitably meant a lot of travel. In 1997, he returned to Ukraine briefly to head up Credit Suisse's new investment banking division in Kiev, before joining Bermuda-based Apollo Fund Management Ltd. in 1998 as director of private equity. In 2000, he jumped to Merrill Lynch, working first in London, then moved back to Toronto in 2001 to establish a technology group.
Grujic also landed plum international assignments. After training in bond trading in Toronto, TD Securities posted him to London in 1998. The job gave him the chance to trade more elaborate products, such as swaptions, which are options that allow parties to exchange a fixed-interest rate security or obligation for a variable-rate one. Grujic and his colleagues also developed computer models for bond pricing. Most bond trading was still done over the phone in those days, but the models could instantly compare the price of, say, a five-year bond with the prices at several different points on the yield curve (from one to 30 years) and determine if the five-year bond was rich or cheap. Hobnobbing was an education, too. "You learn to have dinner with people who've made a billion dollars," Grujic says. In 2000, TD moved Grujic to Tokyo, where he traded even more complex options and derivatives, essentially creating new products.
As things turned out, 1998 was also a pivotal year for electronic trading. The U.S. Securities and Exchange Commission gave the green light to online electronic communication networks, also called alternative trading systems (ATS), to become full-fledged stock exchanges. The enfranchisement of Instinet, Island, Archipelago and Brut meant competition for the NYSE-fast, technologically advanced competition that allowed just about any sizable trader to place orders directly in the market, rather than route them through investment dealers that held seats on the NYSE. There are now dozens of electronic marketplaces in the United States alone, and only about 25% of all American stock trading is routed through the NYSE.
The other gut-wrenching change that opened the door even wider for a geek invasion came in April, 2001, when North American stock exchanges completed the switch from traditional fractional pricing to decimalized pricing. Under fractional pricing, the smallest spread between the bid price on a stock (the highest a buyer is offering to pay) and the ask price (the highest a seller is willing to take) was one-16th of a dollar, or 6.25 cents. Capturing the spread ($125 on even a small retail order of 2,000 shares) had been a reliable source of profit for traditional brokerage firms for decades, and for a mini-invasion of individual day traders in the 1990s. But with decimalization, bid-ask spreads shrank to less than a penny overnight.
Decimalization was a serendipitous development for Tchetvertnykh and Grujic. Through all their early postings, the two men had stayed in touch, and talked a lot about launching their own business together someday. That someday came in 2001, when Merrill Lynch sold almost all of its operations in Canada to CIBC. Tchetvertnykh didn't want to work for a bank, and Grujic was getting restless in Tokyo. As well, both men had recently married, and they didn't want to raise families in hectic international financial capitals. "It was inevitable, so why put it off?" says Grujic of their collaboration.
The duo still weren't sure exactly what the business would be, however. At first they thought of opening a boutique brokerage firm, like Toronto-based GMP Capital, but "the proprietary-trading business model became the most exciting," says Grujic. Although the advent of wafer-thin spreads knocked traditional brokerages for a loop, and wiped out day traders, Tchetvertnykh and Grujic figured there were lucrative niches in the market for new proprietary trading firms. "Banks cannot compete with an innovative, nimble company," says Tchetvertnykh.
So, in 2002, he and Grujic founded Infinium with $1 million of their own money. Off-the-shelf computer hardware was readily available. The hard part was writing the software from scratch-even for two engineers with a decade of high-level experience in global markets.
First, there were automated trading strategies to consider. A lot of them were based on traditional arbitrage between markets, as well as increasingly sophisticated trend-based arbitrage-buying or selling if the price of a security had drifted too far below or above a long-term or short-term trend line. There's also arbitrage between the prices of stocks and the prices of futures, options, swaps, index funds and other derivatives that are based on them, which may take a while to adjust when share prices move.
High-frequency traders have started to assume the role of the traditional market makers as well. In the days of stock exchange trading floors, market makers were individual traders designated by industry regulatory organizations to provide liquidity and maintain an orderly market in specified stocks-if trading sagged and buyers or sellers couldn't find someone who'd accept their order, the market maker was supposed to buy or sell near the latest bid-or-ask prices.
In the modern variation, exchanges and ATSs charge so-called liquidity takers a fee, and give liquidity providers a rebate. In practice, that means that a high-frequency firm might continuously offer to buy or sell a stock. If, say, a traditional brokerage comes in and accepts that offer, it pays the fee. The fees and rebates are tiny-say, 0.003 cents a share for liquidity takers and a rebate of 0.002 cents per share to the liquidity provider (which means the exchange covers its costs). But if you collect rebates on a few million shares a day, they do add up. The same goes for capturing bid-ask spreads, which have been whittled down to fractions of a cent on major stocks, but still can be realized.
Gaming, in the mathematical sense, is also very much a part of high-frequency trading. Other traders, whether they're human beings or algorithms, often trade in patterns. Spot the pattern, and you might be able to trade against them. Of course, everyone in the market is trying to do that, which means continual updates to software are part of the game. Infinium runs about a dozen broad strategies at any given time. "Some might last a week, some might last a year," says Tchetvertnykh. "Our R&D budget is $3 million this year."
Before Infinium could get its systems running, Tchetvertnykh and Grujic had to spend months programming in risk controls, record-keeping functions and tests for compliance with regulations. On any one trade, says Tchetvertnykh, there are dozens of automatic checks within about 20 microseconds.
Speed is so essential that high-frequency trading firms and traditional investment dealers have located computer servers a few feet away from stock exchange trading platforms. Infinium has two servers: one near Alpha Trading Systems Ltd.'s platform in Toronto, and the other near a TSX platform in the suburb of Markham (the exchange has another platform downtown). Transmission time for an order: less than a millisecond.
Yet Tchetvertnykh says "human control" also remains a key component. There are some patterns and anomalies that only savvy traders who monitor the algorithms can spot. One Infinium trader in Toronto who specializes in European stocks and currencies, watches 22 computer screens throughout her working day.
At the beginning, however, Infinium was basically just Tchetvertnykh and Grujic and two other staffers. In 2003, the firm's first full year of operations, revenue totalled just $818,696. Because the U.S. market is so much bigger and more advanced than Canada's, the duo figured they had to expand the business there as soon as possible. Infinium opened a U.S. subsidiary in 2005, and Grujic moved to Marin County, north of Silicon Valley. Good call: Revenue climbed to $13 million that year.
Tchetvertnykh says revenue for 2009 will likely reach $100 million. That's still small: The full-service investment banking divisions of several Big Five Canadian banks each generate more than $1 billion a year in revenue. And privately owned Getco LLC, one of the largest U.S. high-frequency trading firms, employs more than 200 traders, and earned an estimated profit of $400 million (U.S.) in 2008.
rujic and Tchetvertnykh say Infinium's next step is further geographic expansion. The London office, which opened in 2008, gives them a beachhead in Europe, and they're looking at other countries around the world.
Yet they aren't certain how big Infinium will get. Grujic says that "2,000 employees seems to be a natural place we could go." Tchetvertnykh says he'd at least consider going public. "It would give us equity to bring in the best people." He'd also consider selling out-he points out that Citigroup paid $680 million (U.S.) for South Carolina-based Automated Trading Desk LLC in 2007.
Meanwhile, regulators and traditional investment dealers are struggling to assess what high-frequency traders have already done. At a conference in late October, SEC chairman Mary Schapiro said that new rules may be needed "to address new types of market professionals whose activities may not be sufficiently regulated."
High-frequency traders argue that regulators will first have to find problems, and so far, there don't appear to be any. In a study of high-frequency trading in Canada published in September, New York-based Investment Technology Group Inc. made the same argument as the Infinium principals that their kind benefit the market: Bid-ask spreads and share-price volatility in Canadian stock markets is down over the past two years, and order depth-which measures share availability-is up.
Even bank-owned dealers agree with some of that. In October, CIBC published a white paper by six of its senior traders on high-frequency trading and the TSX's rebate program for electronic liquidity providers, which was introduced in 2008. Much of the impact has been obvious: "faster-moving quotes, more bids and offers, more volume, and in some cases, frustration." The CIBC traders also say they believe that high-frequency traders "are not predatory, simply very fast and very good at what they do."
But critics like Thomas Caldwell say that such overall numbers don't tell the whole story. They question whether high-frequency traders are providing "real liquidity" to the market. In fact, argues Caldwell, high-frequency traders are also removing it. How? Large institutional investors know that if they start trying to push through a large block of shares at a certain price-even if the block is broken into many small trades on several ATSs and markets-they can trigger a flood of high-frequency orders that immediately move market prices to the institution's disadvantage. (This is the source of the "frustration" mentioned in the paper by the CIBC traders.)
That's why institutions have flocked to so-called dark pools operated by ATSs such as Instinet, and individual dealers like Goldman Sachs. The pools allow traders to offer prices without publicly revealing their identities and tipping their hand. Caldwell says the best markets for all participants, and for regulators, are "central, open and transparent auction markets," like the old stock exchanges. But the dark pools mean that "all the big orders are now sitting somewhere else."
Risk is also a complex question. The high-frequency traders' supercharged computers haven't blown up markets-yet-but Caldwell says they have blown up individual stocks. Exhibit A: the investment bank Bear Stearns, which folded after its share price plummeted in March, 2008, even though then-SEC chairman Christopher Cox assured the markets that the firm was sound. "Bear Stearns did not commit suicide," says Caldwell. "It was murdered."
Gripes like Caldwell's make Grujic chuckle a bit. "When you listen to vested interests complaining about something, it must be good."