Wall Street: More Machine Now Than Man
Posted by David Gaffen
September 19, 2008, 1:47 pm
Rob Curran reports:
Computers, take a bow. You pulled through a major stress test this week on Wall Street, and helped humans do the same.
Many of the record 10.65 billion shares printed on the New York Stock Exchange composite Thursday were likely routed through the alternative stock exchanges known as dark pools, and executed by computer strategies known as algorithms.
The largest of the dark pools, Goldman Sachs Group’s Sigma X, matched a new record number of shares every day of the week except Monday, peaking at 531 million Thursday. The second-largest dark pool, Credit Suisse’s CrossFinder, saw record volumes, topping 300 million shares on more than one occasion, and Liquidnet, the largest non-bulge-bracket dark pool, saw a record on Wednesday.
computer_art_257_20080919134418.jpg
His computer is taking care of business.
Another sign of dark-pool growth: Liquidnet still plans to go public, in an era when most Wall Street firms are being driven from the stock market. As for algorithmic, or algo, trading, Credit Suisse said its platform for clients has handled more than one billion shares a day, and the same goes for Goldman.
A mere four years ago, semi-automated trading was the preserve of quant-fund geeks, but now, algos are everywhere. When the NYSE adopted the tools this summer, the last opponent to electronic trading had signed up. This week, the NYSE was “frenetic,” said a veteran floor broker, with episodes of “unbelievable stress.” During the crazy openings, the broker said he leaned on algos, because they allowed him to “be in 15 places at once.” By plugging order requirements on quieter stocks into his computer, the broker found he could concentrate his open-outcry energy on the stocks expected to have a boisterous opening.
“I think speed of execution becomes paramount in volatile markets and electronic trading is faster than human trading,” said Dan Mathisson, head of the advanced-execution services algo unit at Credit Suisse.
Some veterans say the ouster of the New York Stock Exchange as a central market hub by a fragmented hodgepodge of electronic venues has created a “Wild West” market structure, facilitating volatility. The evidence suggests, however, there’s no turning back now.
Algorithm is a blanket term covering a range of order types that traders can punch into computers. The idea is to reduce donkey work for traders. Once a clients’ parameters are punched into an algo, it can skip through various market venues and execute in a matter of seconds.
There are risks with depending too heavily on algos: One was exposed by the UAL Corp. fiasco last week, when a five-year-old story about its bankruptcy was accidentally republished. The selling that took place may have triggered some algos to automatically sell. That’s why most traders warn against a Ron Popeilesque “set it and forget it” strategy.
Trickier is the possibility of chasing your tail. If a trader has too many algorithms out there hunting one particular stock, they could end up bidding against themselves, said Jim Gorman, a spokesman for Liquidnet.
While traders need to be aware of the risk, computer trading is indispensable, particularly during these almost unprecedentedly volatile markets. “I think we’re at one of those junctures in time where the ability of a human being to handle the crush or the rush of data requires computer assistance,” said Alfred Berkley, a former president of the Nasdaq and now head of trading-technology firm and dark-pool operator Pipeline Trading Systems. |
