Algorithmic foreign exchange is more than just flavour of the month
Euromoney, is part of the Delinian Group, Delinian Limited, 4 Bouverie Street, London, EC4Y 8AX, Registered in England & Wales, Company number 00954730
Copyright © Delinian Limited and its affiliated companies 2024
Accessibility | Terms of Use | Privacy Policy | Modern Slavery Statement
Foreign Exchange

Algorithmic foreign exchange is more than just flavour of the month

There has been a lot of talk about the use of computerized trading in foreign exchange. Discussions at a recent seminar in London suggest that there is real substance, not just hot air, behind the chat.

It was standing room only at a breakfast briefing held by algorithmic trading provider Progress Software in London on March 7. As is normal for these events, the company had expected some no-shows. On the day, though, even the provision of extra chairs was not enough to prevent a sizeable scrum having to stand at the back of the room.

The audience was made up of buy-side and sell-side participants – some of these were already using algorithmic and program trading, some had looked at them and others were there to see what all the fuss was about.

For a corporate event, the seminar was extremely lively and surprisingly balanced. Several questioners even asked whether algorithms were really the next best thing or just smoke and mirrors. As with all debates, the true answer probably lies somewhere in the middle ground.

Chris Skinner, vice-president of research consultancy TowerGroup, started the proceedings. He said that it was the regulatory and compliance field that was driving a lot of the developments in algorithmic trading, and technology was being increasingly used as the solution.

“Technology is causing competition in the execution space and execution is becoming a commodity. If execution is a commodity, then algorithmic trading gives you the differentiation,” Skinner claimed.

Man feeds dog

Skinner suggested that as a result of the use of algorithms, trading banks of the future would simply consist of a computer, a man and a dog. The computer would be there to handle the transactions, and the dog would be there to stop the man from interfering with its trading programs. The man’s job would be to feed the dog.

John Bates, vice-president, Apama Products, at Progress Software, said there was now definitely an increased uptake and interest in non-equity algorithmic trading. “It’s really starting to happen in FX. Algorithms are now doing anything that traders and market makers might have done in the past,” he said.

Bates tried to provide a definition of what algorithmic actually meant. In the past, the term was used distinctly for those programs designed to get better execution. And although this is still generally the accepted definition in the equity space, Bates said that as far as FX was concerned it covered all forms of program trading.

He added that an algorithmic arms race was now starting to occur. “People really are now focusing on time to market. The first mover gets an advantage. Once the strategy works, it has to be constantly updated.”

Bates also said that there was now widespread talk of reverse engineering taking place, designed to identify when algorithms were active in the market. He said there were rumours doing the rounds of certain proprietary desks arbitraging their own institutions’ models. These types of allegations have also been made in detail to Euromoney, which suggests that the algorithms are not so sophisticated that experienced dealers cannot spot when they are trading and take advantage.

After the seminar, Bates said both reverse engineering and front-running by prop desks were entirely feasible concepts. With regards to the latter, he said that algorithmic trading had its origins as a tool designed to get prop desks better execution, so it was not surprising that some traders were able both to spot their impact and pre-empt their market entry. Bates said the next generation of algorithms would have built-in “genetic modifications”, so that they constantly evolved.

Liquidity is crucial

Much was made during the seminar of the seemingly fragmented nature of the spot FX market. Bates pointed out that to work effectively, algorithms needed either a single liquidity pool or concurrent access to multiple trading venues. “Deep liquidity makes algo use easier,” he said later. “FX doesn’t have the same structure (as equities) trading liquidity pools. That’s why there’s an interest to monitor all the pools and sit on top of them.”

A recent report from the Tabb Group, entitled Outlook on algorithms, also highlighted this fragmentation and pointed towards it as slowing their uptake. “The percentage of FX volume that is traded electronically is growing quite rapidly but is hampered by the fact that it is a fragmented dealer market. Aggregation is the name of the game in the FX space, as it would allow traders to develop smart-routing mechanisms to shrink the disparate market to a single order book,” wrote the report’s author Adam Sussman.

However, this aggregation, rather than the use of algorithms, is causing the so-called liquidity mirage in the FX market and creating situations where various platforms can be “spoofed”. This is proving irksome to certain sell-side traders, but a boon to some of the smarter, more technologically advanced players on the buy side who can spoof the market more frequently and swiftly than was possible in the past.

Jeremy Smart, executive director at Morgan Stanley, pointed out that despite the pitfalls there were numerous benefits to electronic trading. He said market access was now easier and lower prime brokerage was also further helping to open up Reuters and EBS, which he described as the central markets.

Blurring distinctions

Smart said it was for the sell side to adapt to the challenges posed by the new entrants to the market. “Newcomers to FX are forcing the market makers to catch up,” he said. “These guys don’t always have an FX background. They don’t require or need large amounts of capital. These guys can trade two, three or four yards a day in $1 million increments. Often they can be two men and a server, or the market-making part of a large hedge fund.” He added that algorithmic program use was blurring the distinction between the buy and sell side.

According to Smart, the sell side had to learn not to pigeonhole all of its clients and appreciate their differences. By taking a flexible approach, Smart believed that it was possible for both the buy and sell sides to trade profitably.

Looking ahead, all three speakers agreed that multi-asset class trading was fast approaching and that algorithms would have a big part to play in executing strategies in terms of order timing and execution as well as trade identification.

Another theme to emerge throughout the seminar was the importance of transaction cost analysis (TCA). The use of algorithms has made it possible to analyse more fully the true cost of trading. Interestingly, though, even though it seems that plenty of TCA has been carried out on the most efficient way of dealing, it seems little or no research has been carried out on the true cost of designing, implementing and maintaining an algorithmic trading solution. That in itself seems to sum up nicely the somewhat schizophrenic nature of the FX market.

Gift this article