FX market bones up on its maths

David Adams investigates why algorithmic trading is breaking out from its traditional equity stronghold
and penetrating the foreign exchange market, and considers the effect this is likely to have on the sector

There are still plenty of people who cannot believe algorithmic trading will ever be as wide spread in the Foreign Exchange (FX) sector as it already is in the equities market, but the number of FX transactions being executed electronically continues to rise, confounding the sceptics. That rise has not even been checked by increased volatility in FX and the wider markets generally due to the credit crunch, indeed, in a roundabout way, increased volatility may even have contributed to it, as it brings traders using the most sophisticated technology into the asset class in search of alpha. So why does algo trading suit FX? And what effects might it have on that market?

There are basically two sides to algo trading. First there is ‘black box’ rules-based trading, where a machine makes decisions about when to trade based upon a pre-set strategy, brings together the necessary means to do so, and then executes the trade. This is the way many investors now pursue their trading strategies, the way banks automatically hedge risks, and is the driving force behind the rise of algo trading across the financial markets. The second important element is algo execution, which provides market participants with the ability to prove to their clients that they have done all they can to find the best price for a transaction, and thus can act as a competitive differentiator. It helps firms transact in the most economically efficient way, at speed, and is particularly attractive to hedge funds, because it means large volumes of trading can take place in the dark without moving the market.

But in FX there are complicating factors that will mean algo technology will never be used in exactly the same way, or to the same extent, as it is in equities. On the one hand, it appeals to those market participants trading for profit, because the ability to trade in more than one venue offers an opportunity to obtain a more transparent view of the market and prices, making it easier to find liquidity. On the other hand one should not forget the people dealing in FX not for direct profit, but because they require a particular currency to settle a transaction involving another asset class, or as part of a hedging strategy.

Mark Warms is general manager, EMEA, at FXall, one of the main electronic communication networks (ECNs) that enable direct FX market access. He sees the rise of algo trading as just one part of an ongoing move away from traditional business methods in this market. “I think first you have to look back at the FX market as it was,” he says. “As recently as the Millennium 90 to 95 per cent of volumes in this market were traded over the phone, over the counter. That meant that price discovery and transparency were difficult. Then along came electronic trading. Systems like ours allow clients to continue to trade over the counter, but still receive multiple prices from multiple sources.”

This capability introduced changes in the way market participants have to manage risk, particularly on the buy-side. “Once a marketplace allows buy-side traders to connect and trade electronically there’s a shift in control,” says Laurie Berke, senior consultant with the Tabb Group of researchers. “If I’m a money manager, once you offer me the opportunity to access pricing not just with one trader but across multiple venues, it then becomes a challenge for a trader to rest assured that at all times they have accessed all the best prices. If I think about that from a risk standpoint, the risk of me making a mistake has gone up because I can’t be sure I am taking advantage of all the liquidity available. I’m taking control of where I go to trade, but as I do that, I begin to take on the responsibilities for those decisions.”

Hedge funds and exchanges
Harrel Smith, head of product strategy at trading software specialists, Portware, says his company is aware of the increased interest in FX algo trading, which he believes was driven in the first instance by hedge funds looking for alpha. “We’ve seen
some doors opening for hedge funds that had previously been closed,” he points out. “Now you have fixed income electronic
trading that was previously closed to the buy-side open to them as well. I think the growth of algo in FX is part of a broader
market-wide initiative to bring buy-side liquidity to electronic trading. It’s about the decision to allow buy-side to participate more
on those platforms.”

Increased electronic trading certainly seems to be having some positive effects for those trading exchanges that have adapted
fast to the new world. In early May CME Group was able to announce healthy increases in trading volumes, including a 41 per cent increase in FX volumes in the year up to April 2008. Derek Sammann, managing director, and head of FX products at CME Group, credits the electronic platforms available for future and options business, and the fact that more market participants are now seeking to generate returns from FX – particularly since the market became more volatile in the second half of 2007 – for these healthy figures. At the same time, CME’s co-location service, which allows customers to co-locate application servers with the CME matching engine server has helped reduce latency. “Algo trading is probably about 40 per cent of our business right now, so that still leaves a lot of volume that is latency-sensitive,” Sammann explains. “For these people it’s about how much information they can gather. Can customers take all the streams of information in, and execute on aggregated price? Technology is closing these gaps, and making it cheaper and more efficient for people to make the right trading decisions. The choices are being made at a more sophisticated level, as people apply the same techniques that you see in equities.”

In an algo-driven world it also seems likely that the ability to prove best execution will become more important. Tabb Group’s Berke suggests that while there is no direct regulatory imperative to obtain best execution in FX, regulations like the EU’s Markets in Financial Instruments Directive (MiFID) have effectively forced many firms to develop the technical infrastructure to support an audit trail that proves it has been achieved. “Regulation in equities has created a culture of measurement and quantification, and [market participants] can also use this as a competitive advantage, using the same electronic tracking tools,” she explains. “So there is a culture of best execution being created.” In many cases the nature of the FX market means it will be relatively straightforward to demonstrate best execution, but it won’t always be possible, in more complex structured derivatives, for example.

Technical challenges
Managing a world where algo trading is becoming more commonplace also presents some extra technical challenges, and vendors are only too eager to try and help. Wall Street Systems aims to help market participants improve the way they cope with the sudden bursts in trading volumes that occur when automatic trading systems are triggered. “What many banks are struggling with is getting hit by 1,500 trades all at once,” explains Rick Schumacher, head of products at Wall Street Systems. “They’re asking ‘how can I understand the impact of that on my position and my risk’?” “Let’s say you’re a market maker, and you get 5,000 trades in, backlogged for 15 minutes. You have no idea what your position is. You maybe can guess, from your order flows, but the problem is usually, when trading, that its highest peak is when rates are at their most volatile. That’s when you, as a market maker, or a trader, need the best information possible. That’s the area of most risk.” At the same time, banks may also face problems further downstream where trade processing backlogs build up, raising the spectre of latency concerns and execution problems.

In-house developments
The spread of algo trading has created new opportunities for those financial institutions that have developed their own high-end algo-based systems. Last year, Credit Suisse began work on an electronic Foreign Exchange (eFX) solution to offer to its clients, as well as its own traders, building on the success of its Advanced Execution Services (AES) algo trading platform. This spring the resulting solution won the FST Award 2008 for the Best Use of IT in Wholesale and Investment Banking. In December 2007, the bank also started trading structured FX options on its Merlin internet trading platform, offering clients the ability to structure, price and execute exotic option transactions online.

“We didn’t rush into eFX when it was all about high volumes and vanilla flow,” says Ian Green, managing director and head of algorithmic and electronic trading for FX at Credit Suisse. “Our background is in doing the innovative, added value things. Credit Suisse did well establishing the AES brand, and we extended that into FX, initially with some internal users. There are other people doing some things that sound similar, but they actually offer a less wide choice of algos and generally are about trading on the bank’s stream. What we do is quite different – people are transacting without going through the bank’s market making books.”

Green’s colleague Jonathan Wykes, head of AES FX in Europe, even claims that the system can improve communications between the bank and its clients. “AES is very strategy-oriented,” he says. “With a lot of rival vanilla eFX services, the clients take the product but don’t talk to [the bank or service provider]. With this there’s an ongoing conversation. Traditionally what’s happened wherever you’ve had more technology is that things have moved away from salespeople. It’s quite the reverse with Merlin and AES.”

Latency
This is part of the reason why, while Green will accept that of course latency remains an important issue, it isn’t necessarily the be-all and end-all for electronic trading. “Latency is important, but with us the actual design of the services is even more important. We want to make the service that we provide convenient to the client and relevant to what they want. User design matters, and Straight Through Processing (STP) matters.” (Interestingly, this echoes recent research carried out by CME Group looking at attitudes among banks, money managers and other FX market participants to eFX, which revealed they were far more concerned about counterparty risk and settlement risk than about latency).

“When you look at latency across different venues and portals, there’s not all that much difference,” says Wykes. “Should you spend millions making it faster, or should you be spending that money on making better, more flexible products?”

It is possible that eFX is driving a fundamental change in the way the FX market operates, with more transactions inevitably moving over to those trading venues that emerge from the current period of consolidation, leaving the old sell-side model behind. Wykes believes one of the most significant effects of the spread of eFX may be a more systematic approach, what one might even call a more rational approach, to trading venue selection, which is likely to lead to further consolidation among ECNs. “I think you’ll see some of the weaker ECNs fall by the wayside,” he says. At the same time, of course, dealers will need to work on making sure they can still offer value to clients who can now find so much information, so quickly and directly, when the circumstances suit.

It appears that the ‘new’ FX market is just as tough as the old one, says CME Group’s Derek Sammann. “There were concerns about how deep this liquidity is when a crisis hits,” he says. “Will these electronic models be able to cope? I think if you look at what has happened recently, with the current volatility problems, the answer is a resounding ‘yes’. Not only has liquidity not decreased, it has actually increased. It shows the stability and the robustness of a lot of the electronic trading in the market.”

Nor do there seem to be many good reasons why Luddite traders or bankers should fear over-reliance on technology or worry about resilience. “I don’t think the dangers are that great,” says FXall’s Warms. “Individual banks or customer systems may break down occasionally, but people become more efficient at having redundancy built-in, and everybody has backups. I don’t think there are huge risks associated like that. In general, I think algo trading, when you look at the banks managing their risks, has been a huge boon for the marketplace.”

Credit Suisse’s Green agrees: “With our Merlin system, you’d think if it’s going to be priced, booked and hedged electronically there might be more scope for errors, but we have seen very few mistakes,” he says. “It’s partly an STP initiative, and STP has reduced risk in pretty much every area it’s been used.”

The tide of technological change may, of course, result in some lower-level trading positions being replaced by automated algo trading systems but these will still need human beings to programme them, control them and think up innovative ways to harness their power.

So what next for algos in FX? With technical risks reducing, capabilities and market volatility increasing, and the use of algos spreading throughout the market, it promises to be an interesting time. “The use of algos is established now for internal processes,” says Green. “The pertinent question is what can you do for your clients to bring the benefits of these algo capabilities to them?”
The next few years will give us the answer to that question.

more features

abc home