EMAIL DETAILS

Algorithmic trading: Market keeps machines in their place
By Jeremy Grant
Published: September 27 2010 17:38 | Last updated: September 27 2010
17:38
Ever since the “flash crash” of May 6 in the US, when the Dow Jones
index plunged by an eye-watering 1,000 points in about 20 minutes, the
machines that increasingly move markets have come under the spotlight.

Specifically, regulators question what role algorithmic trading had in
exacerbating the huge price dislocations of that day, when the market
also bounced back up not long after it had fallen.

Yet computer algorithms – or “algos” – are by no means used solely in
equities markets. They are just as pervasive in options, futures and
foreign exchange. In FX, about half of market activity is driven by
some kind of algorithmic trading.

Specialist technology providers are scrambling to roll out the latest
sophisticated systems designed to help FX traders use algos more
efficiently.

In September StreamBase, a US-based company, teamed up with
MarketFactory, a technology company providing algorithmic FX trading
applications, to give hedge funds, banks and proprietary trading firms
access to market data more easily.

Companies such as this know FX is growing faster than cash equities
markets as a trading opportunity for algorithmically inclined traders.
Indeed the Bank for International Settlements says the FX market grew
by 20 per cent globally in the three years to April this year. Some
exchanges that offer FX are doing even better: CME Group’s FX volumes
grew 94 per cent in that period.

So could a “flash crash” happen in FX – and does the rapid growth of
FX provide extra cause for concern?

Most industry experts start with the “never say never” caveat. But
they all point out that the FX market structure has unique
characteristics that mean it is less likely than some people fear.

One is that it is less fragmented across trading venues and pools of
liquidity than equities markets.

In the US equities markets, one of the issues that regulators are
focusing on is the fact that trading was spread across 14 exchanges,
and the numerous different types of trading platform may have
exacerbated price movements as traders switched back and forth from
various venues trying to get out of their trades.

Dave Rutter, chief executive of Icap electronic broking, says: “In the
equities markets there are thousand of securities that are traded
thinly. If you take a look at our top [currency] pairs like the euro/
dollar or euro/yen, these are very deep liquidity pools with thousands
of participants.”

He says algorithmic trading has been “an important part of our market
for the last five years.

“We’ve seen that the algorithmic firms have added liquidity and added
to the depth of the market.”

James Dalton, head of FX algorithms at Citi, says the FX market is
nonetheless becoming more complex. There is no single source of real-
time market data or volume “so you have to model it yourself”.

Moreover, algorithmic trading is a generic term that refers to
electronic market-making (done by a handful of top-tier banks that
dominate the business); “automatic technical trading” (often done by
hedge funds); statistical arbitrage players such as high-frequency
traders, who are simply, seeking to take advantage of speed; and the
emergence of “execution algorithms” used by traders to get large
volumes done but parcelled out into smaller trades.

Mr Dalton says “stat arb” players make money when markets are volatile
but this can bring “a mix of benefits and dangers”, he says.

“That’s why more banks are starting to invest in building [execution
systems] for their customers to use, so they can navigate those waters
a little more effectively.”

However he also believes a “flash crash” in FX is unlikely.

“One that would bring the market to its knees? No,” he says. “The
market’s just way too wired. A lot of the algo business doesn’t sit on
particular position for very long.”

In the meantime, progress is being made in automating many mundane
post-trade processes that are designed to give traders a better sense
of who they are trading with, and the risks of default – known as
“counterparty risk”.

Portware, an algorithmic trading technology company, said in May it
had integrated one of its trading products with post-trade systems
operated by Traiana, a company that specialises in middle- and back-
office processing of over-the-counter derivatives.

This would allow Portware’s clients to “streamline post-trade
processing and settlements with counterparties, resulting in lower
operational risk and reduced trading costs”, the companies said.