Since the infamous stock market “flash crash” of May 6, high-frequency, or algorithmic, trading has been unwillingly dragged into the political and regulatory limelight.
So far, however, attention has focused on the role of these high-speed traders in the equity market. Outside the glare of that publicity, it is less well known that on May 7, FX trading volumes reached records, straining the plumbing of these markets.
Some participants argue these strains were partially caused by algorithmic, or algo, traders.
Exactly how much of this can be attributed to algo trading is unclear. However, there is no question that high-frequency traders are a fast-increasing force in FX markets, which is sparking a fierce debate as to their value to the market.
On Tuesday, the Bank for International Settlements reported that average daily turnover in the FX market has jumped 20 per cent in the past three years to $4,000bn a day. Its survey was taken in April, so missed the May spike, which related to the eurozone sovereign debt crisis.
The BIS-reported gains were led by a near 50 per cent leap in spot trading – deals for immediate delivery – to $1,500bn a day. This jump was powered by increased activity from “other financial institutions”, a group that includes hedge funds, pension funds, some banks, mutual funds, insurance companies and central banks. This will also include algos.
While all categories of “other” could have increased their trading, it is likely a significant proportion was driven by algo traders, who favour the deep, liquid spot markets and particularly currency pairs such as eurodollar and dollar-yen, which between them account for 42 per cent of all currency trading.
The question for the FX market is whether high-frequency dealers improve the market by adding liquidity, or whether they are instead merely price takers who contribute little.
“Algos have been demonised, but they’re an important part of the growth story,” says David Rutter chief executive of Icap Electronic Broking, which runs EBS, the main FX interbank trading platform. “What we’ve found is that they add pressure at each price point so that instead of getting big price gaps on shocking news, trade is more orderly.
“With FX, there are a lot of other flows such as global trade, so there is good underlying liquidity that the algos can enhance.”
Algos initially appeared in FX markets almost a decade ago, attracted by the deep liquidity and increasing use of electronic trading. They were generally welcomed, particularly by banks looking to build their prime brokerage businesses. However many banks soon grew disenchanted when they found the fast-moving shops were profiting from banks’ own slow systems by exploiting brief, tiny price differences between rival platforms.
Some banks went as far as ejecting offenders from their platforms but banks’ views have since become more nuanced. They have generally reached an accommodation, helped by technological improvements which make it easier to monitor client dealings and offer client-specific prices.
“The facts are that algos have made the markets more efficient and have helped ensure there’s one virtual price,” says Jeff Feig, global head of G10 FX at Citigroup. “They do cause banks to be smarter and we’ve had to work harder to be more efficient, but that’s ultimately to the advantage of the end user.
“I think that to some extent, algos have pushed banks and the result has been enhanced transparency and increased liquidity.”
Algos mean many different things in the FX market. While high-frequency traders are the best known – typified by one senior banker as “five smart guys in a room in New Jersey,” – banks are increasingly adept at developing their own algorithms to make their internal FX deals more efficient. These “internalisation” trades too will have provided a boost to the BIS numbers.
Most players say algos are now a fact of life in currency markets.
Unlike the equity market, which is split into hundreds of stocks, they believe the FX world’s focus on a relatively small number of currency pairs means it would be far harder for a single group of participants to move the market significantly, intentionally or otherwise, as some watchers fear happened during the “flash crash”.
“Also trading can happen anywhere there’s an electronic execution system and a volatile market,” says Alan Bozian a former FX banker and now chief executive of CLS Bank, the FX settlement system. “The question is, which markets adapt well and I don’t think it’s necessarily the stock market.”
FX markets have proved generally good at adapting. Systems such as CLS, introduced years before the financial crisis, have helped minimise settlement risk and since May, participants have been working again to improve their processing systems to cope with increased volume.
Significantly, for a market that is very much built around a hub of big banks, the BIS report showed that, for the first time, interaction of the main banks with “other” financial institutions overtook trading between themselves.
This could be a pointer to the market of the future, where banks are likely to remain the hub, but as much for their trade processing abilities as for their liquidity.
This would allow the winners to build profitable volume without taking on huge trading risks – suiting the current regulatory mood.
“The banks want to continue being the price providers, but they’re getting much more interested in the infrastructure and improving that,” says Mr Bozian. This evolution is likely to apply to high-frequency trading too.
Mr Rutter believes algos are only in their “late teens” in terms of development. “The early algo trading was about super-fast dealing and chasing inefficiencies. That’s largely gone,” he says.
“Now its about math and science being thrown at the market – there’s a rich pool of data and I think we’ll see algos evolve so its not just about milliseconds, but about longer-term predictive math.”
"FT" and "Financial Times" are trademarks of the Financial Times. |
