Cynics might wonder whether there's intelligent life to be found
on Wall Street, but within the financial community there's plenty of
life in artificial intelligence. AI, as it's commonly known, is the
programming of computers to approximate the workings of the human
brain. Far more than mere number crunching, the focus is on
sophisticated tools like neural networks that simulate thought
processes. Neural networks allow computers to discern subtle
patterns as they winnow mounds of data and to actually learn as they
see new patterns emerge.
In today's volatile markets, the capacity to learn makes AI
especially intriguing to portfolio managers. It would be naive to
expect a black box that could automatically pinpoint money-making
opportunities, says Andrew Lo, a professor at MIT's Sloan School of
Management. But AI has "proven potential to identify temporary
inefficiencies, which can then be used to beat the market." Lo is
heading a recently launched Research Project in Computational
Finance to look into the ways computer intelligence can be applied
Most financial organizations are experimenting with AI, but
only a few have actually put it to use. Citibank's currency traders,
for example, are working with a neural network. Merrill Lynch uses
an AI program to help price 1 million corporate bonds each day.
Risk assessment. A couple of money managers are producing
impressive results. Since its founding in 1986, LBS Capital
Management has been using AI to "decide how much risk there is in
the market and to create portfolios that are either aggressive or
defensive," states LBS President Walter Loick. With nearly $300
million in individual and institutional money under management, LBS
has outpaced the Standard & Poor's 500-stock index by 24 percent
over the past six years; its average annual gain has been 16.3
percent against the S&P's 14.0 percent. The LBS portfolio of
midsize company shares has done even better. Over the past five
years, the S&P Midcap index has gained an average of 21.6
percent annually, while the comparable LBS investment fund has
compounded by 35.8 percent.
The Safety Harbor, Fla., firm, one of several companies
helping to fund Lo's research, uses a computer model that decides
when to be in stocks and when to be out. A "buy" went into effect at
the end of February. The LBS computers have recently focused on
interest-sensitive shares such as insurers Chubb and General Re,
Pittsburgh's Mellon Bank and bond trader Salomon Brothers. Battered
medical stocks like Mylan Laboratories, U.S. HealthCare and United
Healthcare have, meanwhile, been added to the LBS midcap portfolio.
The previous buy signal, which came near the end of last
October, correctly anticipated the stock market's strong ascent
during the final quarter of 1992. At year's end, however, perception
of rising risk produced a "sell." And, in fact, the market did take
a sharp, though brief, tumble in mid-February just before President
Clinton unveiled his economic program.
Brad Lewis, who runs the Fidelity family's Disciplined Equity
fund, also has done well with AI stock-picking techniques. Each day, his neural web checks some 2,000 stocks for 11
variables, such as projected earnings and debt levels. The object is
to find out "inefficiencies in stock prices, a process the human
brain can't perform very well with 2,000 companies."
Disciplined Equity gets a top five-star ranking from
Morningstar, the mutual fund research firm. The fund averaged a 15.2
percent annual return in the past three years, compared with 8.8
percent for the S&P 500. And in 1992, Disciplined Equity
registered a 13.3 percent gain, compared with 7.6 percent for the
S&P. "Even more remarkable, it didn't have a losing quarter in
1992's rotational market," notes Morningstar analyst Tom Desmond.
"Investors who have not yet discovered this fund should take a
closer look, because it's a gem."
Lewis, who also runs the Fidelity Stock Selector, a
two-year-old sibling of Disciplined Equities, has become more
defensive recently, loading up on the shares of utilities like Long
Island Lighting and oil stocks like Mobil and Chevron. "This looks
like a time when capital preservation is more important than capital
gains," he says.
That AI is hardly a precision instrument is apparent from the
mixed signals coming from the systems used by LBS and Fidelity.
Human judgment is still a constant--until someone comes up with a
neural network to reconcile the disparities.
|Sign up to
receive our weekly tip sheet to what's new at
to U.S.News & World Report magazine and receive 4
FREE ISSUES!. |