Commentary

John D'Antona Jr.
Traders Magazine Online News

CEO CHAT: Tina Byles Williams, FIS Group

Investment veteran is on the lookout for talented, lesser known managers in frontier and emerging markets.

Traders Poll

Are you concerned about foreign ownership of a U.S. stock exchange?



Free Site Registration

August 31, 2004

The Age of the Algorithm

By John A. Byrne

Also in this article

  • The Age of the Algorithm

Is Another Disaster Just Around the Corner?

Some brushes with trading disaster during the last bull market are unforgettable. Next month will mark the anniversary of one of these near apocalyptic moments: Black Monday, October 19, 1987. That's when the Dow Jones Industrial Average sank 508 points - a huge 22.6 percent plunge back in those days.

Then there was the Long Term Capital Management (LTCM) fiasco - a $1.2 trillion hedge fund, run by some of the brightest minds on Wall Street. It cracked up in 1998 when Russia defaulted on many of its international debt obligations.

These two separate meltdowns have at least one thing in common: Each was propelled in part by complex computerized, or program trading, models. Some analysts suggest that Black Monday was triggered by massive selling the previous week, which led computer models to initiate a universal wave of sell orders. The LTCM explosion is believed to have occurred because the fund's models had never been programmed for a Russian default.

Program trading is now more common than a decade ago. Is this a cause for alarm? Is another electronic trading blowup now more likely?

Fortunately, there are more computerized safeguards, according to Dr. Michel Debiche, President and CEO of Quantia Capital Management, a statistical equity arbitrage hedge fund in Princeton, N.J. "There is a web of automated quantitative strategies sitting out there ready to pounce on any anomaly in the market, providing a major stabilizing force," he tells Traders Magazine.

The cases of Black Monday and LTCM are extreme. But they do underline the potential perils of computers that take control in trading while human traders have their hands off the wheel.

"These two events suggest that model-driven trading, left unchecked, can cause short-term liquidity crises," according to Celent Communications.

Yet, despite the demonstrated problems, the trend seems to be toward more and more program and algorithmic models. Some of the reasons are all well known: the lack of available shares - or liquidity - at various price points and the reputed efficiency of computerized executions. Another reason, sometimes overlooked, is the demand for conflict-free independent research. This is forcing some money management firms to find separate services for executions and research.

"Because program trading desks provide execution only and are not tied to soft-dollar or research products, these desks benefit from the unbundling of trading and other brokerage products," according to a report published late last year by CIBC World Markets.

However, not all sellside execution firms are ready to jump on this program trading bandwagon. "We have taken a different route," says Doug Rivelli, who runs program trading at Weeden & Co. with Michael Mook and Dion Albanese. "It has a mathematical overlay but a much more behavioral approach."

At Weeden, a hybrid-risk trading model includes a team of human portfolio strategists - who analyse pre-trades and reams of data - and other pros who augment the value of computer models.

"What we have found is that pure risk bids are getting more expensive, and more importantly, are not very efficient at maximixing execution performance," Rivelli says.

Unlike other program houses, Weeden's human traders might work, for example, 10 to 20 percent of the client's order, with the rest handled algorithmically.