Howard Phillips is getting e-mail messages and telephone calls from investment managers, bankers and individual investors and says he is a little surprised by the queries.
Phillips has never worked for a financial institution, nor had he ever before done any work related to the stock market. But that changed when he discovered an unexpected confluence of his disciplines and market forecasting -- a new field called 'financial engineering.' Phillips has a doctoral degree in electrical engineering and computer science; his master's is in nuclear engineering. He teaches digital electronics and computer hardware design at the University of North Carolina at Charlotte. These days, for an annual fee of $25,000, Phillips tells his clients where the prices of stocks, bonds and mutual funds are headed.Phillips got into his new line of work by chance. He was a vice president at Semiconductor Research Corp., in Research Triangle Park, N.C. -- part of a manufacturing sciences research team designing a control system that could anticipate temperature changes and adjust thermostats accordingly.
An acquaintance at a brokerage firm heard about the research and asked Phillips if it could also forecast financial peaks and valleys. It couldn't, but Phillips adapted his research and came up with a new program that crunches financial data three different ways. First, it applies a 25-year-old computer science concept called 'learn and optimize' to adapt a trading range to market movements. Then it applies a 100-year-old mathematical concept called 'functional analysis,' which Phillips describes as 'a way to do what a human does when he looks at a graph.' Finally, the program applies 'filtering capability,' which separates trends (order) from random price movements (disorder); it has been common in telecommunications and electrical engineering for at least 50 years, says Phillips. 'I make no claims that these are new inventions,' he says. 'I just brought those three things together.' Phillips uses a soft-sell; before asking for money he lets prospective clients try his service for a few months. 'It sells itself,' he says. Computers crunch data and fax results to clients to give them a head start on the next day's trading.More than a year ago, Phillips started posting part of his financial engineering forecasts on an Internet site. His first page includes his credo: 'Forecasting accuracy can be consistently higher than 60% using financial engineering.' Phillips says his accuracy typically tops 80%. The site lists Fidelity Investments Select Funds, showing Phillips' forecast for the net asset value of each and some other data, including a plus-sign or a blank by each fund, which indicates whether his latest prediction was right or wrong. Lately, the list has had a majority of plus-signs.
The Internet address (http://www.coe.uncc.edu/~hphillip/) wasn't easily found by Internet cruisers until counter-cultures came together under a new name with a simple address. Yahoo, an Internet index with links to thousands of sites, (http://www.yahoo. com), came across Phillips' site and put it on the Yahoo Mutual Funds home page. Phillips' site is Yahoo's only link under the heading 'Analysis.' The Internet Audit Bureau, which counts visits to Internet sites -- known as 'hits' -- estimates that Phillips' site is hit 6,000 times a month. Some days, Phillips says, he receives more than 200 e-mail messages from curious visitors. The site gives a direct link to Phillips' e-mail address and lists his telephone and fax numbers. He got so bogged down trying to answer queries that he has posted a disclaimer saying he can't respond to every inquiry. Most answers can be found, Phillips says, in a 70-page book he wrote -- the graphs, charts and complicated explanations make the information unsuitable for his Web site. With book in hand, and a daily visit to his site, investors can spot changes in his forecasts, and that's the best time to buy a fund or a stock, he says. He posts forecasts daily, for the next trading day. But Phillips tells people not to buy the $25 book if the library has it. Writing it was purely a not-for-profit effort, he explains. With the shipping charges, he barely breaks even. So he urges investors to try the local library first. 'Don't buy something you can get from the library for free,' he says. But, cautions Phillips, investors must use information in the book if they expect to maximize profits and minimize losses. The Web site, of course, is also a not-for-profit venture -- but it has turned into a dandy tool for marketing his forecasts to well-heeled investors and firms. For the moment, though, Phillips isn't looking for new clients, he says; teaching and research demands have limited his time to crunch data. The site is just a means of backing up his claims of forecasting accuracy. If small investors manage to turn a profit by using the information on it, he says, 'that's great.' Phillips' site isn't full of fancy graphics, so it takes little time to download each page -- and that's just what first-time users should do. Phillips is a researcher, not a writer, and his text is loaded with jargon, instructions and disclaimers. The wise thing for a novice to do is print out the whole thing, study it -- and then get the book. Note: This revised reprint, prepared by D. H. Phillips, is a condensed and edited version of the orignial copyrighted Dow Jones News article by K. Haines. Kathryn Haines regularly visits mutual fund and market analysis sites on the Internet and reports on them for Money Management Alert readers. Her address is dj.mma@ccmail.dowjones.com UNCC encourages and invites corporate sponsorship of graduate students carrying out research on new and advanced applications of forecasting theory. Howard Phillips is at The University of North Carolina at Charlotte, telephone 704-547-4828.