Stacking the cards


It looks like it should be a simple question of arithmetic, but predicting a company's earnings is more like handicapping horse races: part system, part intuition.

"It's an art, not a science," said Bruce Smith, an analyst with Merrill Lynch.

Here are a few considerations analysts throw into the mix:

David Readerman, a Montgomery Securities analyst, said high-tech earnings estimates can be divided into two categories. One group--composed of companies that make, say, semiconductors and disk drives--relies on manufactured components. The production of these components can be measured to gauge the upcoming demand. But other companies such as software makers have no physical parts, so there's no supply movement that can be monitored for changes. This second group, obviously enough, is harder to gauge.

Estimates from other analysts can also sometimes be a factor in an analyst's calculations, according to Eugene Glazer, a Dean Witter analyst.

"The most recent estimate change can be the most important. This has a high influence because all other professionals may wonder why the change and call the company. This sometimes leads to other revisions later on," Glazer said.

Michael Kwatinetz, a Deutsche Morgan Grenfell analyst, offered this anecdote: "A company I follow came to our [investment] conference in December, and based on their talk, we thought their quarter looked stronger than previously expected and raised our estimates. The stock went up. Then analysts, listening to the same presentation by the same company at the next conference, thought the company was not going to do as well as they hoped and downgraded their estimates. The stock went down...So everything is in the eyes of the beholder."