Algorithms Overseas
Electronic trading is on the rise in almost every market across the globe, and, not surprisingly, so is algorithmic trading. But, in the most general terms, European adoption of algorithms lags about two to three years behind their adoption in the United States, and adoption in Asia is perhaps a year further behind Europe, experts say. According to Westborough, Mass.-based TABB Group, about 13 percent of all order flow in the U.S. passes through algorithms; the total for Europe, Asia and Canada combined is just 3 percent.
Given the unique requirements of each market, however, this disparity should not come as a surprise. It would be naive to assume that it is possible to simply apply algorithms that have been tried and tested in the U.S. market to say, the South Korean market.
Even as the globalization of markets and standardization of electronic communications protocols have softened once-hard barriers between national markets, the patterns of algorithm use and adoption overseas cannot mimic those in the United States. Regulatory and structural differences still persist among the U.S., European and Asian markets, and among national markets on the continents. As a result (and as always), before entering a new market, it is critical to know the territory.
Algorithm-Friendly Markets
There are some basic factors that make a market attractive for algorithms. Most important, the main exchange in the market needs to match orders electronically. This is the reason that there is more algorithmic trading on the Nasdaq than on the New York Stock Exchange, which receives orders electronically but typically matches them on the trading floor, relates Adam Sussman, a consultant with TABB Group and author of the industry analyst's report on global trading.
Second, a firm quote environment is needed. Mere indications of interest (IOI), even if they are electronically disseminated, are not sufficiently certain enough to support algorithms. "What if you do not get a response? What is the appropriate amount of time to wait?" Sussman asks. "A computer does not deal with that kind of uncertainty as well as a human would."
Liquidity also is important. The efficiency of an algorithm is minimized when there are fewer choices to be made and there are wide gaps between the bid and asked price on shares.
Less significant but still a factor, a standard protocol such as FIX is helpful to algorithmic growth, Sussman notes. While the FIX protocol is not explicitly necessary to facilitate algorithmic trading, there is a correlation between markets that use FIX and the volume of algorithmic trading, he suggests.
Lost in Translation
Additionally, some algorithms are more appropriate for non-U.S. markets than others. The two most basic and popular algorithms - volume weighted average price (VWAP), and the price-priority or implementation-shortfall (IS) algorithms - are concepts that can be applied almost universally, according to Sussman. More opportunistic algorithm types, however, vary in their efficacy from market to market.
The first algorithms in wide use in the U.S. were so-called "smart routing" programs that were intended to overcome the market fragmentation resulting from the creation of electronic communications networks (ECNs), such as Instinet, Archipelago and Island, that had risen up to compete with Nasdaq and the NYSE. Such algorithms do not have much of a role in most European and Asian markets because exchanges in those markets control a vast proportion of total liquidity and there is not a substantial off-market or ECN community, according to Octavio Marenzi, founder and CEO of Boston-based Celent Communications.
In some cases, this is a matter of law, Marenzi notes. In France, for example, there is a "concentration rule" that mandates that all trades must go through the Euronext exchange, he says.
However, Europe is due for a structural overhaul that is expected to have effects similar to those seen after the adoption of the order-handling rule of 1997, which spawned ECNs in the United States. The Markets in Financial Instruments Directive (MiFID), scheduled to be implemented by November 2007, will standardize market practices across Europe, essentially creating a best-execution standard that will obligate brokers crossing shares internally to report that they have met the best available price in the market and to post quotes publicly. [For more on MiFID, see page 31.]
This quote data would have to be disseminated, and if there is one thing every algorithm needs, it is a healthy data stream. That eventually could lead to the kind of scenario that is well-known in the United States - market centers rebate brokers for contributing quotes and thus attract order flow, improving the climate for the creation of alternative trading venues and for algorithms, according to TABB Group's Sussman. In Europe, this kind of development is probably two to three years away, he asserts.
"MiFID could cause the same kind of fragmentation that Reg NMS and the order-handling rule caused," Sussman predicts. "Then the prices in all those markets change rapidly, and it becomes too much of a pain for a human to work those orders, so you automate the process of looking for the best bid and offer in that marketplace," he explains.
Ary Khatchikian, president and cofounder of Portware, a vendor that supplies algorithms to brokerages and institutions, agrees with Sussman's assessment. "MiFID could change the condition of smart order routing not being so useful where countries have concentration rules," he says.
Tweet This