Automated orders push trading capacity limits
Tuesday June 12, 4:48 pm ET
By Jonathan Keehner
NEW YORK (Reuters) - As automated trading bombards equities markets, the networks connecting U.S. market centers are being tested by overwhelming volume.
Widespread adoption of computerized trading has fueled this growth, as automation not only increases the volume of trades but also the number of electronic messages sent by traders to determine price quotes and volume.
But fears now exist that an unusual surge in trading -- for example, the type brought on by an unexpected event -- could paralyze U.S. market centers. An explosion in the use of algorithms, forecast to account for half of all equity trading by 2010, is raising concerns.
"Certain events could trigger a massive reaction," said Sang Lee of financial services consultant Aite Group.
In algorithmic trading, computers carry out complex strategies, often according to shifts in market momentum. They automatically send messages to market centers to test prices and volume and then convey various buy, sell or cancel orders.
For each order actually completed, algorithms may generate many times that number of messages or quotes. They are relayed between broker-dealers and market centers at lightning speed using computer protocol languages.
Quotes received by the NYSE Group, parent of the New York Stock Exchange, averaged above 90 million per day in 2006, according to Aite Group. A decade earlier that daily average was less than 800,000.
"With algorithmic trading there is exponential growth in overall messaging volume," said Lee. "A lot of orders suddenly get canceled as part of the strategy. That taxes the overall capacity of execution venues."
Growth in messaging volume dwarfs that of actual trades. NYSE Group, now part of NYSE Euronext (NYSE:NYX - News), receives about 20 times more messages than it did five years ago, while the amount of executed trades is about six times greater, according to Aite Group data.
"STUFF IS GOING TO BREAK"
The New York Stock Exchange and other marketplaces already experienced major delays when besieged by massive volume during a market slide earlier this year.
On February 27, trades on the Big Board had to be executed after the official close as the S&P 500 experienced its biggest one-day drop in years, falling nearly 3.5 percent.
The NYSE has expanded its capabilities since the February delay, said NYSE Euronext chief executive John Thain at the Reuters Exchanges and Trading Summit last month.
But he could not rule out the Big Board or another exchange being paralyzed by message traffic.
"We don't get paid for messages, we get paid for trades," Thain said. "So nobody has infinite message traffic capacity."
Mark Palmer, general manager of the Apama Division of Progress Software, which makes algorithmic trading platforms, said growth in automated trading has stretched the capacity of U.S. equities markets.
"Unfortunately, we see it on a weekly basis," said Palmer, when asked about trading delays related to volume. "Those that hit the media are the big ones, but it happens all the time. Stuff is going to break and it's going to break a lot."
A trade getting snarled by automation isn't necessarily new. A growing use of computer programmed trading, among other factors, was widely blamed for the stock market crash of 1987. On October 19, or Black Monday, the Dow Jones Industrial Average (DJI:^DJI - News) had its largest ever one-day percentage drop -- more than 22 percent -- on unprecedented volume.
Of the 2,257 NYSE-listed stocks on Black Monday, there were 195 trading delays and halts and 280 more on the following day, according to a report by the U.S. General Accounting Office, which investigated the crash at the request of Congress.
RISK TO MARKET OR LESS NIMBLE PLAYERS
What's different today is the capacity of computers to pump out massive volumes of price checks and orders in less than the blink of an eye.
For individual, institutional or retail investors trying to get out of a position quickly during a market collapse, any disruption could mean they get a much worse price than they might have if it were running efficiently.
Experts are divided about the gravity of the situation. Some argue that new demand will drive needed capacity.
Adam Sussman, of consultant Tabb Group, says that if a market center proves incapable of handling growing message traffic, another player will eventually take its place.
"I don't see that as a risk to the overall market," said Sussman. "I see that as a risk for less technologically nimble players within the market."
But Aite Group's Lee says the preponderance of algorithmic trading could exacerbate a market downturn.
"Algorithms are built to look for anomalies," said Lee.
Indeed, a problem in calculating the Dow Jones industrial average may have triggered a sharp rise in volume that led to the Big Board's February delay, which took place after a sell-off in China precipitated the U.S. market plunge.
"Now if something happens in the Chinese stock market everything goes haywire here," said Lee. "Assuming global algorithmic adoption rates increase, then what?"
But Sussman of the Tabb Group says there are a sufficient number of players in the market to stabilize even a catastrophic decline.
"There are enough people on different sides of the risk-reward scenario that they will balance each other out," he said. "Some will be on the losing side of that volatility but there should be enough liquidity to soak it up."
Dan Mathisson, who heads advanced execution services at Credit Suisse Group (VTX:CSGN.VX - News), finds the fears unfounded.
"Algorithms have been operating for a long time," he said. "People who don't like electronic trading have been waiting for years for something to happen. You can never prove there won't be a problem but it's not like there weren't problems before."
(Additional reporting by Anupreeta Das)
|
