ZIFFPAGE TITLEWhy Human Traders areBy Larry Dignan | Posted 2005-03-07 Email Print
The exchange wants to automate its trading systems. But many clients prefer dealing with human specialists. How John Thain will try to reach both objectives.Better (and Worse)">
Why Human Traders are Better (and Worse)
With the hybrid system installed, NYSE's challenge will be to find the balance between productivity and stability. Here's how those two worlds collide.
On Dec. 10, 2004, Pfizer announced that its Celebrex arthritis drug leads to heart attacks. The stock traded as high as $25.95 and as low as $21.99; more than 289 million shares changed hands.
Before the NYSE opened, Pfizer traded on the electronic exchanges at around $22 a share. With 100 people standing around his post and 37 million sell orders, Schafer delayed Pfizer's open until after 10 a.m., or about 45 minutes after the company made its announcement, so traders could evaluate the news.
Schafer opened with a price of $23.52, about $1 higher than the price on the electronic exchanges.
As Schafer explains, "$22.52 to $23.52 may not seem like much, but if you're the money manager who dumped 7 million shares at $22.52, you have to explain why you lost $7 million." He says the Pfizer incident shows the benefits of the NYSE's format.
Kevin O'Hara, chief administrative officer for electronic net Archipelago, disagrees. He says the electronic exchanges did all the heavy lifting and found a stable price well before Pfizer even traded on the NYSE. "When it was an extreme period for Pfizer, the NYSE shut down its post so it couldn't trade," says O'Hara, who adds that the NYSE "was keying off the electronic price."
How that Pfizer trade will happen in a hybrid market is a hot topic for specialist firms such as LaBranche & Co. Chairman and CEO Michael LaBranche has spent his last two earnings conference calls fielding questions about a hybrid system at the NYSE. On a Jan. 21 call, LaBranche said automation should boost productivity as specialists rely less on keystrokesas many as 20,000 to 30,000 a dayand more on programs that react to and interpret a stock's "behavioral patterns." To shepherd these projects, Britz says the NYSE is spending much of its time making sure it identifies what traders and specialists need and building their requirements into the system.
Once floor traders, specialists and other market participants have their say in new technologies, changes will be rolled out slowly and evaluated. For example, Direct+ can handle more than the 159 million shares it traded on an average day in January, but is limited so the NYSE can study its impact on volatility.
"We implement in a very measured way, even with good design and QA [quality assurance]; we're integrating a series of related subsystems and we have to make sure there aren't unintended consequences," Britz says.
Its devotion to human involvement in the auctions and big trades means the exchange mostly writes its own software.
"By and large, we do all of own cooking," Britz says. "We get what we want, and there aren't any places in the world that do what we do and on the scale we do. Obviously, if we could go out and buy cheaper and faster, we would, but frequently we don't have the option."
Institutional and individual investors will have to wait until at least March 2006 to see what Britz has cooked up; and whether Thain can keep his board, specialists and seat owners all happy with the resulting broth of men and machines.