Warren Harding. Calvin Coolidge. Herbert Hoover. Jimmy Carter.

George W. Bush?

The stock market fell during those president's terms and all four became one-term presidents.

That precedent, at least according to Wall Street folklore, doesn't bode well for Bush. The Standard & Poor's 500-stock index has fallen 19 percent since he was inaugurated in January 2001.

Whether stocks go up or down during the first four months of an election year is also supposed to predict who wins or loses, another bad omen for the president. Others on Wall Street say you can pick the president based on the market's performance from New Year's Eve to election eve, or between the end of the second political convention and the election.

Most political scientists dismiss such theories as voodoo. But, they insist, if you put enough academics in front of enough computers, eventually one of them will come up with a sophisticated economic formula that will predict elections.

This year's election pits the folklorists against these formula writers, or -- as they prefer to be called -- economic modelers.

Most of the models call for a Bush victory based on such factors as the advantage of being the incumbent, polls, economic growth statistics and the employment picture.

In the stock market, though, the stars align in the opposite direction, pointing to the inauguration of Democrat John F. Kerry. The Dow Jones industrial average and the S&P 500 fell during the first four months of 2004 and are still down for the year.

Jeff Hirsch, editor of the Stock Trader's Almanac, cites the correlation between market performance in the first four months of the year and election results. "Of the six times the market has declined in the first four months of the year, five times the incumbent parties were ousted," he said. The incumbent party, rather than the incumbent president, is tracked to take into account people who aren't running for reelection.

At the end of April, the S&P was off 0.4 percent and the Dow was down 2.2 percent. The Nasdaq Stock Market composite index, down 4 percent through April 30, is ignored because the Nasdaq market hasn't been around long enough to have a track record for predicting elections.

Hirsch also tracks each president's performance from Election Day to Election Day, arguing that the market begins responding to a president the day he's chosen rather than waiting for the inauguration. On that basis, Harding, Coolidge, Hoover and Carter were destined to lose and Bush is in big trouble.

That is also the case if you happen to believe that what happens from New Year's Eve to election eve really determines who wins. As of Friday, the Dow was down 4.7 percent this year and the S&P off 2.3 percent.

But Bush supporters should not despair. Hirsch contends that a fall rally bodes well for incumbents, as does Sam Burns, senior equity analyst at Ned Davis Research Inc. in Venice, Fla.

It would take a 400-point rally by the Dow, but just a 16-point advance by the S&P, to get those indexes back into the positive territory by Election Day.

Both men look for rallies starting after the second political convention -- the Republican convention this year ends Sept. 2 -- through the day before the election.

"If you think the market will predict the election, wait until after the conventions," Burns said. "If it takes off and looks happy, it looks likely the incumbent party will win."

Going back to the election of 1900, the Ned Davis Research data show that when the Dow has fallen between the conventions and the elections, the incumbent party has lost seven out of 10 times. When the market has rallied right before the election, the incumbent party has kept the White House 13 of 16 times.

All that is challenged by researchers at the University of Iowa. Studying elections is a specialty in Iowa, my home state, where the first-in-the-nation political caucuses every four years are the biggest things that ever happen.

"Taking one measure and finding a correlation between that and the presidential election is not a sufficiently rigorous way to study the problem," business professor Thomas A. Rietz said. "It's not as simple a relationship as, 'The stock market went up, therefore the incumbent got reelected.' "

His colleague in the political science department, Michael Lewis-Beck, tries to predict the presidential election with an economic model based on polls, economic growth, job creation and the incumbency advantage.

These models are not perfect, either. The quality of new jobs -- a big issue this year -- isn't worked into most of them. Neither is foreign policy, which also looms large over the November election.

Instead of unemployment, job creation is a key factor in the model developed by Lewis-Beck and Charles Tien of Hunter College in New York. As Democrats like to point out, Bush gets bad grades on that score.

"What you see is that under the current administration there have been fewer jobs created than in any other" since World War II, Lewis-Beck said.

However, when you throw in healthy economic growth in the first half of the year -- 4.2 percent -- and the magical incumbency factor, the Iowa model gives Bush 51 percent of the popular vote. But since the calculations can be off by as much as 1.5 percentage points, the election is declared too close to call by Tien and Lewis-Beck.

Pioneered by economist Ray C. Fair of Yale University, presidential election models have been created by a number of economic forecasters and consultants. They have a spotty record, though, political experts say. In the 2000 election, many economic modelers missed, predicting a big victory by Al Gore.

Fair, the author of "Predicting Presidential Elections and Other Things," called the last election for Gore, who won the popular vote, but said accurately that it would be close. In April he predicted that Bush would win this time with almost 59 percent of the popular vote.

Another widely cited economic model, created by the economic consulting firm Global Insight Inc., gave Bush 56 percent of the popular vote in a report July l.

Rietz uses a market to predict elections, but it's his school's own market. The University of Iowa runs a political futures market in which traders, using real money, can buy futures contracts on either Bush or Kerry to win the election.

It's based on the "efficient market" theory of prices. The idea is that markets provide the most accurate estimate of what a company or a commodity or a politician is worth, because they reflect the combined decisions of all the people who are knowledgeable enough about the subject to risk money.

The advantage of a market over polls, Rietz argues, is that polls reflect the opinions of votes picked at random -- whether they care about the election or not. The political futures market is based on the inputs of people who at least think they know what they are doing.

"We've seen Bush ahead most of the time, but when Kerry picked Edwards it brought them up to a much closer race," Rietz said.

Based on last week's closing prices, the Iowa futures market was split 50.1 for Kerry to 49.9 for Bush. In other words: too close to call.

Jerry Knight's e-mail is knightj@washpost.com.

A number of economic models based on stock market performance are purported to predict election results.