Here is the bad news about ultrahigh-speed trading systems: Financial advisers and individual investors will never be able to keep up.
The good news is that in some ways, it doesn't really matter.
“It's certainly not a fair fight when it's individuals versus supercomputers,” said Bob Van Valzah, who took part in a panel discussion at the Securities Industry and Financial Markets Association technology exposition last week in New York.
“But the good news for everyone, including advisers, is that this all leads to more-efficient markets,” said Mr. Van Valzah, director of product marketing with 29West Inc., a unit of Informatica Corp. that markets superfast order-messaging technology.
Because most individuals and advisers tend to adopt a buy-and-hold approach or trade infrequently, the volatility that can change securities prices hundreds of times a second tends not to be an issue for most long-term investors.
For institutions, however, including the mutual funds through which most investors make equity purchases, superfast trading is a way of life. They participate in the roiling markets every day and are constantly subject to staggering losses and gains within timeframes as short as a few thousandths of a second.
Although it doesn't affect them directly, advisers should understand how speed-sensitive trading works, because doing so affords a glimpse of where financial technology is headed and what regulators will be up against.
“A one-microsecond loss can add up to 5,000 lost trades,” said panel member Donal Byrne, chief executive of Corvil Ltd., an Irish company that provides systems for high-performance trading networks that manage latency, which on Wall Street means the time between when an order is entered electronically and when it is executed. The firm is part of a technology segment that has emerged over the past few years to optimize trading networks and shrink latency and data loss during trading.
UNDER A MICROSCOPE
Mr. Byrne described how a European exchange approached his company in 2007 to help detect microbursts of rapid trading. To illustrate the phenomenon, he displayed a graph showing a horizontal line representing order traffic at one-second intervals.
A dramatically different view appeared when the same traffic was viewed at five-millisecond intervals, which is akin to looking at trading through a microscope under higher magnification.
The second slide had more in common with a side-on profile of the Himalayas, showing innumerable crests and troughs representing the latency and data loss associated with microbursts.
“Seeing [the pattern] with modern technology is pretty straightforward; dealing with it, on the other hand, can prove more problematic,” Mr. Byrne said.
Currently, trading entities tend to set what are called “latency and loss objectives.”
A typical goal would be to get 99.999% of orders executed during a market's busiest single second.
The SIFMA panel discussed several ways of approaching that and similar targets, all of them expensive in terms of additional bandwidth and hardware, and all coming with trade-offs.
In recent benchmark testing, Cisco Systems Inc. and Informatica found that some of the fastest hardware and network architecture achieved latencies as low as 660 nanoseconds (that is 660 billionths of a second) on an optical network, resulting in a throughput rate of 21.1 million messages per second. Think of that volume as buy and sell orders coming out of just one trading entity, and you begin to see just how fast the world's markets can move these days.
Such speed, of course, creates the potential for a market disaster, as exemplified by the “flash crash.”
“Going back to May 6 — people didn't even know if they were in or out of the market for two weeks,” said Dave Malik, director of solutions architecture at Cisco Systems.
Identifying the underlying causes of the flash crash is crucial in order to avoid a repeat, said Peter Lankford, founder and director of the Securities Technology Analysis Center LLC.
“The markets are performing at levels we never thought possible, and in general, what the industry understands [about what happened in May] is only the tip of the iceberg,” he said, noting with other panelists that another flash crash is entirely possible.
Given the risks, maybe it is time for the Securities and Exchange Commission to recruit even more computer geeks.
E-mail Davis D. Janowski at djanowski@investmentnews.com.