Foreign Policies

From a niche strategy to one of the principle channels for executing equity trades, algorithmic trading has risen in the past few years from an industry buzzword to a staple of the modern trading environment. The foreign exchange (FX) market, however, has been slower to adopt algo, despite the rapid development of electronic trading platforms in the space.

"The movement toward electronic FX trading (e-FX) has been growing for the past seven or eight years," says Tim Cartledge, managing director of FX at Barclays Capital. Historically, he says, the FX market's fragmentation and the relative illiquidity in some segments have meant that this method of execution has not found the foothold it has elsewhere.

There are also so-called "invisible pools" or "dark pools" of liquidity, which are privately negotiated trades that are not necessarily actively advertised to the marketplace. "FX has not gone toward the exchange model as equities and futures have. As there is so much liquidity on the OTC market, it's not appropriate to use the same models," says Cartledge.

Some of the barriers to widespread adoption remain in place, says David Ogg, CEO of LavaFX, which recently released a time-slice algorithm for FX. "There is no national best bid and offer or reliable, aggregated source of market data," he says. Finding an algorithm to operate effectively in the FX market is not simply a question of reworking one from equities space. "It requires recognizing that the market is different," he says.

"The key to get algo moving is to get depth of liquidity, and then overlay the algorithms on top of that," Cartledge says. "If you haven't got that base level of liquidity, you have to make it." BarCap did just that when in late September it unveiled an algo feature, PowerFill, on its BARX e-FX platform, which offers its clients the algorithms that it uses for its proprietary FX trading. As with equities, Cartledge says that initially the key benefits are "transparency and the ability to manage transaction costs." The PowerFill display on the BARX user interface allows traders to gauge how well their orders are being absorbed and this enables them to minimize their trades' impact on the market.

Bank of America (BofA) released its own trade weighted average price (TWAP) algorithm in the same week as BarCap's PowerFill, but Scott Freedman, head of BofA's FX electronic trading services (ETS), says the development of FX algorithms should lend heavily from the equities space. "We think there are a lot of synergies between the two," says Freeman.

Part of the challenge is assessing the landscape, according to Freeman, who says there is no single obstacle put in place by the market structure that prevents a broader uptake of algorithmic trading. "The major barrier is just coming up with the models," he says. The pace of technology over the past few years and the growing experience of developers of algorithms have driven the systems into new asset classes. The acquisition in January of Financial Labs, a specialist algo trading and pricing company, gave BofA a team of mathematicians and astrophysicists, which accelerated its FX algo program.

One point of contention, according to Freeman, is how to define algo trading. "If you go to eight different banks and eight different clients, you'll get 16 different answers as to what algo is and what it isn't," he says. He says there are no technology barriers in place to an algorithm that simply chops up an order into fragments and executes it. "The ability to trade on a benchmark has been around for a long time. That's not how we see an algorithm."

One group hoping to accelerate the shift in methods of trading FX is FXMarketSpace, a joint venture between the Chicago Mercantile Exchange (CME) and Reuters. The platform, set to launch in the first quarter of 2007, seeks to become the first centrally cleared OTC FX marketplace and to provide a structure closer to the exchange model that the market has so far been lacking. "What we see is a natural evolutionary path in FX trading," says Bryan Hunter, COO of FXMarketSpace.

From the earlier single-bank portals, through multi-bank and electronic crossing networks (ECNs), FX trading has largely been controlled by the banking community, says Hunter. "But the bank-to-customer paradigm is changing," he says. Buy-side traders are increasing in sophistication, which is driving them to look for new platforms and venues on which to deal. "Market-maker, market-taker is not a paradigm that makes sense," he says.

FXMarketSpace uses the CME's existing infrastructure for clearing, as well as a central limit order book and matching engine on its Globex platform. Executable spot price values will be fed from the venue to give participants real-time pricing and volume data. Detractors say the FX space is already fragmented and that the same liquidity will now be spread onto yet another platform. One bank official, who requests anonymity, says FXMarketSpace is along the same vein as the many ECNs already in the arena, and not necessarily a venue on which his bank would be happy to put liquidity. The official says that market making in this area would be "a losing proposition."

"We don't think we'll be taking anyone's liquidity," says Hunter, who says the platform will open up the FX market to new participants and increase the levels of trading by existing players, particularly by areas of the buy side that already use black-box and systematic methods.

Cartledge says e-FX has not only changed the way that buy-side clients have worked, it has altered their relationship with the market. "Traders have found trading FX electronically puts them at a distance from the market. They have less control than when they were dealing in the interbank market directly," he says. Algorithmic trading should enable them to regain that control, he says.

One recurring suggestion that arises from discussions on the subject of highly developed FX algorithmic trading is that the technology could open up cross-border, cross-asset trades. For example, an equity order could be placed on an overseas exchange, and tied into that could be an FX spot trade in the local currency.

"There are certain things that are achievable," says LavaFX's Ogg. There are differences, such as settlement dates, in the respective market structures that throw up hurdles. "It's feasible to save a few pennies on execution, but to lose money overall," he says.

Cartledge says the idea of combining equity and FX algorithms suffers from a mismatch in the size of the two markets. Since equity orders tend to be much smaller, he says, it is not economical to cover FX at the equity ticket level. "Instead, the FX risk from equity algorithms will need some degree of consolidation before being hedged," he says. However, there is some mileage in the possibility of hybridizing algorithms for FX and batches of equities.

Some say FX algo needs to learn to walk before it starts sprinting toward a multi-asset goal. "The FX market is doing a bit of catch-up," Ogg says. "Participants need to be aware of the rest of the world and to take advantage of what's happening in other markets." He says the FX market is likely to reach levels of adoption comparable to other asset classes. "At the moment it's only niche because there's not a lot of it," he says.

Others are equally enthusiastic about the future. "We're having a revolution," Cartledge says. "The old ways of trading FX are being revised." For his part, BofA's Freeman is less bullish about assertions that algo will come to dominate the FX market as it has in equities. "I don't predict the future," he says. "But there is a real demand for it now."