As a leading strategist for global banks and electronics firms, Chris Skinner has for decades watched financial fads come and go. The latest thing everyone's talking about is called algorithmic trading. But this time, even Skinner is convinced: Algorithmic trading, he avers, truly is going to be the “Next Big Play.”
Next Big Plays, says Skinner, who runs Balatro, a U.K.-based think tank on the future of financial services, don't come along often. The Last Big Play, Skinner says, sprang from the creation of the derivatives market. That happened almost a decade ago. These days, derivatives trading is a US $12-trillion affair.
With well over a quarter of U.S. and European equity trading now powered by algorithmic trading systems, Skinner says it's time to face the facts: Algorithmic trading ? in which traders rely on computers and software designed to automate not just the worldwide execution of their trades, but in some instances much of the thinking behind those trades ? is reshaping world financial markets in astonishing ways, and with astonishing speed.
“The markets are no longer bounded by execution venue, exchanges, geographies, brokers and dealers,” says Skinner. “Trading moves to where liquidity is highest.”
The implications of these developments for investors ? large-, medium- and small-scale alike ? are enormous. According to a study by Boston-based TowerGroup, a division of MasterCard Worldwide, algorithmic trading will push direct market access trading (trading without broker intervention) to 38% of total buy-side flow by 2008. In 2006, 27% of U.S. hedge fund trading, and 16% of trading by large institutional investors, flowed through algorithms. A recent IBM study suggests that around 40% of the trades made on the London Stock Exchange may now originate from algorithmic trading systems.
And many of these aren't ordinary trades. The development of automated technology has allowed traders to input scores of real-time market variables and reams of statistical data drawn from multiple historical databases marketed by data powerhouses like Bloomberg and newly-merged Thomson-Reuters. Algorithmic trading has in turn used this automation to propel complex, cross-asset class trading strategies that combine instruments ranging from equities to foreign exchange. The resulting trades can be simultaneously conducted in many differing types of markets across the globe.
But getting in on this game is expensive. The pioneers in the field were Credit Suisse and Goldman Sachs, both of which were able to muster billion-dollar investments in technology and data sources to better service their clients, especially hedge funds.
Investors with access to the technology can boast impressive results. According to Tower Group studies cited by Chris Skinner, traders with access to increasingly advanced “algo” systems report yields in the 12% annual range and volatility as low as 12% between 1998 and 2005 ? a big contrast with the general equity markets that delivered under 4% yield and up to 18% volatility in this period. In a similar vein, Andy Webb, editorial director of U.K.-based Automated Trader Magazine, told Canadian Business Online (CBO) that transactional cost savings achieved by large volume traders using algorithms to place trades are “pretty good and plenty of hedge funds out there are doing 20% to 30% yield on automated trading.” Seen this way, algorithmic systems look like money-making machines for those with vast asset bases under management ? although retail investors may soon gain access as well.
After TD Newcrest ? Canada's No. 1 equity block trader ? struck an arrangement with Goldman Sachs Algorithmic Trading last November, Ray Tucker, managing director of institutional equities at TD, traced his decision to client demand: “Our clients had been talking for a couple of years about the buzz in the U.S.,” Tucker says. TD says the clients currently interested in algo trading are those that routinely execute large volume trades, like pension funds and hedge funds. But as the bank said in an e-mail message to CBO, its discount brokerage is also exploring “delivering appropriate access to algorithmic trading to retail investors to meet a small but growing interest in the strategy.”
TD's decision to rely on Goldman Sachs to develop what Tucker calls a “Canadianized” algorithmic trading package speaks volumes about the systematic advantages algorithmic trading gives those with the deepest pockets. Even with TD's relative heft in Canadian equities trading, Tucker says it couldn't go it alone: “We started down the path of building our own product, and then we realized we couldn't compete with the likes of Goldman Sachs, which has 500 or so programmers working on its algorithmic systems.”
Traditional brokers aren't alone in finding these new waters challenging. Stock exchange insiders like Joe Rosen, until last year a senior technology manager for the New York Stock Exchange, say algorithmic trading also poses problems for traditional exchanges like the TSE.
In a market where automated trading systems are easily able to move amongst scores of exchanges and electronic trading venues to place orders ? or slices of orders ? around the globe, “the traditional exchanges are at a technological disadvantage,” Rosen argues.
The need for the speed to capitalize on momentary margins offered by shifting currency valuations, split-second market lags or differences in transaction costs, is changing the trading landscape. For one, it's driving large trading volumes toward specialty trading operations like Kansas City-based Bats Trading, which in recent months has emerged as the third largest U.S. stock trader by volume, thanks to highly-specialized computer software innovations affording ultra-fast, ultra-responsive trading (which attracts algorithmic business) and deep discounts in trading prices. “This is probably not a good thing for the old-line exchanges,” observes Rosen.
Apart from anything else, Rosen notes, the traditional exchanges will have a tough time competing technologically with powerhouse specialists like Bats. Instead, some exchanges like the Nasdaq have been buying up some of the pioneering electronic trading platforms that threatened to poach large volumes of trading activity from mainstream exchanges.
At the TSX, Rob Fotheringham, VP of trading, acknowledges no threat. That's because algorithmic traders, he argues, will always favour highly liquid markets like the TSX.
“As a trader you always want to go to the liquidity. You want to reduce your footprint in the market, to reduce market impact,” says Fotheringham. “So you go to where there's greatest liquidity. Our aim is to remain the centrepoint of liquidity. To do that we've invested heavily in the core technologies and hardware that facilitate algorithmic trading. We're redesigning our core trading engine. That'll take us up to world class efficiency.”
As brokers and exchanges scramble to adjust, those who face the biggest risks from algorithmic trading may very possibly be small-scale investors ? those most reliant on market stability, good governance and fair access to information and technology.
“Many of the algorithmic trading strategies include risk strategies untested by market downturns,” warns Balatro's Chris Skinner. “A lot of the traditional market stabilizers are being taken out” by the trading speed and fluidity enabled by algorithmic trading, he adds.
The biggest stabilizer of all, he notes, is perhaps the role of local and national financial regulators.
But in the new era of algorithmic trading, in which trading technologies and strategies are closely-held competitive secrets fiercely protected from scrutiny, Skinner says regulators will have a tough time gaining sufficient information to “understand what they are regulating.” Especially if it's happening in 10 or 20 jurisdictions simultaneously. It all starts to make Enron's famously complex trading strategies look simple, he adds.
After all, Skinner notes, if major Canadian banks find it necessary to outsource their algorithmic trading expertise because it's too expensive to go it alone, how likely is it that our notoriously parsimonious financial regulators ? with their laughably small technology budgets ? will be able to ensure the world's increasingly fleet-footed algo-experts always play fair?