ePredict.com Brings Political-Style Polls To Investing

By PETER LOFTUS - Dow Jones News

NEW YORK -- In late December, shares in online direct marketer MyPoints.com Inc. (MYPT) were riding high, having soared tenfold in the four months since their initial public offering.

With MyPoints.com's stock price on a precarious perch, a new financial Web site called ePredict.com asked its users Dec. 29 if they thought the stock would outperform the Standard & Poor's 500 index over the next three months.

A majority of ePredict's users answered that MyPoints would underperform the S&P; 500, prompting the Web site to issue a "strong sell" rating for the stock. Sure enough, the shares lost half of their value by Feb. 2, and even after a partial recovery they were trading this week at nearly 30% off their Dec. 29 closing price. Meanwhile, the S&P; 500 dropped just 5.4% in the same period.

Call it populist investing. In this presidential election year, ePredict.com has brought political-style polling to the art of picking stocks. Unveiled last week, the Web site surveys its users each day about their views on various stocks and industry sectors. The poll results determine the Web site's stock recommendations, which are e-mailed to registered users each morning before the markets open.

ePredict calls the method "systematic analysis" and views it as a viable supplement - or alternative - to the more traditional investing tools of technical and fundamental analysis.

But is it accurate? The Web site claims its poll-based stock ratings were correct 65% of the time in preliminary testing between October and February. In recommending technology stocks, ePredict claims a near-perfect accuracy rate, at 94%. The site has about 1,000 registered users and ePredict expects that base to grow quickly.

Scott Pederson, ePredict's founder, president and chief executive, said he got the idea for using polls as an investment tool about 10 years ago, but couldn't find an appropriate medium for it until the Internet blossomed.

"I was just thinking, 'If election polls are so accurate, and if billions of dollars of advertising money are moving based on 2,000 households surveyed for the Nielsen T.V. ratings, wouldn't it be interesting to survey active investors?"' Pederson said. Epredict defines an active investor as someone who trades at least once a month.

The key to using polls as a strategy, Pederson said, is to ask investors not only where they think a stock is heading, but also whether they plan to buy or sell the stock. He compares this to asking likely voters whom they'll vote for, in addition to polling nonvoters.

Hardened Wall Street types might scoff at the notion of using opinion polls to make investment decisions. But ePredict sees the method as a natural offspring in the marriage between the individual investor and the Internet.

"Main Street's more involved in investing," said John Kaminski, Epredict's vice president of marketing and business development. "The individual investor is really becoming much more of a reality in the marketplace. With the advent of the Internet and all the investing tools on the Internet, a lot of people are managing their own money."

ePredict appears to be more of a bull than a bear. Last week, the Web site issued strong buy recommendation for 13 stocks, including Advanced Micro Devices Inc. (AMD) and Vicinity Corp. (VCNT). It had only two strong sell recommendations, for Garden.com Inc. (GDEN) and Outback Steakhouse Inc. (OSSI).

ePredict isn't always right. After Mattel Inc.'s (MAT) embattled chairman and chief executive, Jill Barad, resigned Feb. 3, ePredict.com asked its users whether they thought Mattel shares would rise or fall over the next month. A majority predicted the stock would climb, resulting in a "strong buy" recommendation. But the stock, which closed at 10 1/2 the day Barad stepped down, went nowhere, closing at 10 1/4 on March 3. It closed Tuesday at 9 7/8, down 3/8 for the day.

And at least one investment expert is skeptical of ePredict's systematic analysis. Gur Huberman, professor of finance and economics at Columbia University, said there's a key difference between political opinion polls and investing polls. In a political poll, voters presumably tell the truth in saying which candidate they favor, partly because they want to influence other voters to vote the same way once the poll results are published, Huberman said.

But an investor may have an incentive to lie to a pollster, he said. For example, let's say an investor wants to buy shares in International Business Machines Corp. (IBM), but thinks the stock is too pricey. She might lie to a pollster and say she thinks the stock price will fall, hoping to influence others to sell IBM, thus driving down the price and making it more affordable for her to buy.

Conversely, if the investor has already built a position in IBM stock and doesn't intend to buy any more shares, she might lie and tell a pollster that she's likely to buy more, hoping to influence others to buy, Huberman said. If others buy, the stock price would rise, lifting the value of her portfolio.

"At no cost," Huberman said with a laugh, "you can ask me what will happen to IBM and I can say it's going to go up 20% over the next three months. So what does that mean?" For the record, Huberman is a firm believer in buying and holding stocks for the long-term, and having a diversified portfolio.

ePredict, formed last year with seed funding, is now seeking venture-capital investments, said Pederson, who has 18 years of sales and marketing experience.

In the meantime, ePredict is generating revenue by selling advertising on its Web site. It's also trying to line up deals to distribute its content on other Web sites and media properties.

Ultimately, ePredict wants to put its money where its mouth is - by starting a mutual fund based on investor polling, Pederson said.