Algo FX: the battle for market share

BOSTON - Uptake of algorithmic trading in the foreign exchange market is expected to more than double in the next two years, according to an impact note from Boston-based consulting company Aite Group. However, analysts argue that development of this segment is not without its hurdles.

The consultancy said last week that algorithmic trading will soar from its current uptake of 7% of global FX trading to 15% by 2010. In its note, it said growth to date has been spurred by hedge funds, with proprietary trading shops and currency overlay managers also facilitating the rising trend.

"Activity from the leveraged hedge funds community has contributed to the rising uptake in algorithmic trading," says Pritpal Gill, Citi's Singapore-based head of Asian FX and FX options trading.

The report said algorithmic trading would not have achieved its current level of uptake without the emergence of electronic communications networks (ECNs), such as Hotspot, which enable trading firms with live executable two-sided streaming quotes. The consultancy calculates leading ECNs clock a combined average trade volume of around $400 billion a day.

"An increasing proportion of flow on spot broking system EBS as well as on ECNs comes from participants who are not traditional market-making banks," says Ian Green, London-based head of algorithmic and electronic trading for FX at Credit Suisse. "As usage of these channels increases, users find that larger sizes need to be worked intelligently, and algorithms can help."

Green says access to adequate liquidity by using aggregators and expertise in working orders by using algorithms are needed to provide a credible alternative to the banks' click-and-deal platforms, which still dominate the FX market.

In addition to this, the improvement in the electronic access available to venues such as EBS has led to a dramatic change in the way banks use them for risk management and opportunistic price-taking, says Green. "In pure volume terms, this has been the dominating factor in the growth of FX algorithms to date."

Electronic trading groups from broker-dealers have also intensified growth in the algorithmic FX market. Participants such as Credit Suisse view FX as an ideal market through which to implement execution algorithms because of market fragmentation, liquidity levels and a decreasing average trade size, according to Aite.

While algo trading is set to rise, its increasing adoption is not without potential pitfalls, according to Sang Lee, managing director of Aite Group in Boston.

The first is cost, with banks looking to charge an extra $2 to $3 per million for access to generic execution algorithms. This rises again to as much as $6 or $7 per million to use more sophisticated, customisable execution algorithms.

"Price increases might be justified by the value delivered to the customer, but will be capped due the growing number of algo product vendors," notes a London-based head of FX. "As a result of a growing vendor group, the market will determine the appropriate price per million over time, based on the underlying execution savings achieved."

Furthermore, the increased participation of proprietary trading desks and actively trading hedge funds has made speed of execution an issue in the FX market. According to the report, latency problems can arise when traders want to capture market data from several locations.

"Depending on the distance of the various potential execution venues, a certain amount of latency exists, based on difference in distance as well as the existing IT infrastructure of various execution venues," says Lee.

"When in New York, I see London prices that are about 40 milliseconds out of date," says James Sinclair, chief executive of financial technology firm MarketFactory in New York. "If I try to hit a price, it will be another 40 milliseconds before my order gets to London and even longer before I find out whether I was successful. Even prices from Chicago are seven milliseconds out of date when I see them in New York."

One solution to this problem is for institutions to adopt software such as MarketFactory's Whisperer (FX Week, March 31). The software, which can be installed at market data centres close to various execution venues, reduces latency issues by streaming market data to the client in real time.

"There is an opportunity cost if I miss a price," said Sinclair. "Whisperer solves this problem by using algorithms to provide an institution with an improved view of the order book at local and distant venues."

Meanwhile, there is also the possibility of increased latency arbitrage as foreign exchange algorithmic trading grows. "With more people trading FX algos, there will naturally be more traders engaging in latency arbitrage," comments Fergal Walsh, global co-head of FX e-trading at Citi, based in New York. "However, as a result of heightened competition in the market, the margins to be gained through such practices will decrease with time."

Banks should continue to invest in the best infrastructure and technology for price distribution and market data consumption, to minimise the latency effect, Walsh says. "Co-location with the right exchanges and the position of banks' pricing engines are becoming more and more significant as the trend for algorithmic trading in FX increases."

Aite Group interviewed 12 actively trading asset managers, hedge funds and proprietary trading firms to glean information on the market for foreign exchange algorithmic trading.