Discuss as:

Political picks? Ask again later

— Political prediction markets may be technically more accurate than polls, but they can go through the same ups and downs that polls do. Now that the trend line has had a few days to settle down, it looks as if the Republicans as well as the Democrats got a mini-bounce after their respective conventions - but nothing near the swings that the polls recorded.

Whether you're backing the GOP's John McCain or the Democrats' Barack Obama, you can find a fresh prediction that will warm your heart.

Republicans can take encouragement from the latest projections based on state-by-state electoral votes, which are the only votes that count in the end. This week, researchers led by Sheldon Jacobson at the University of Illinois at Urbana-Champaign reported that their mathematical model gives the edge to McCain. A "neutral" prediction puts McCain's electoral-vote count at 271, beating out Obama's tally of 267. (That's barely enough to win.)

A similar analysis was served up by the University of Washington's Darryl Holman at Hominid Views. The current forecast at Electoral-vote.com also favors McCain, although it leaves 34 votes up for grabs - enough of a margin to throw victory into doubt.

These models draw heavily upon state-by-state polling, and that means there's plenty of time for the forecasts to change. So what's the verdict from other models that supposedly do a better job at long-range forecasting? The margin has tightened just a bit on the political prediction markets since our last status report, but so far the Democrats are still on top.

The GOP shares in winner-take-all trading on the Iowa Electronic Markets topped off at 47 cents on Sept. 12 and have since settled down to 42 cents. The Democrats' shares bottomed out at 53.1 cents on the 12th and are at 59.2 cents today. (To refresh your memory, each share pays $1 in November if the presidential candidate wins, and zero if he loses.)

Yale economist Ray Fair uses yet another long-range mathematical model, based on economic statistics for the quarters leading up to the election. His latest analysis, issued at the end of July, sets the eventual Republican vote share at 48.5 percent.

Since then, of course, we've had not only the conventions but McCain's surprise pick of Alaska Gov. Sarah Palin as his running mate. We're also having a pretty significant economic crisis, by the way.

Fair said the economic situation in particular could make things worse for the GOP.

"It's probably the case that the growth rate [for the year's third quarter] is going to be smaller than the figure I used - in fact, it could be negative," he said. If third-quarter growth is less than the 2.8 percent Fair projected, the GOP's vote-share figure would suffer.

On the other hand, Fair can't account for factors such as Palin's selection - or, for that matter, McCain's "not-your-typical-Republican" image.

"It is an intangible, and it could be a significant one," Fair said. "If he can distance himself from the current administration, it's to his advantage to do so." And that's just what he's trying to do.

In any case, a projected 51.5-48.5 vote split is within the 2.5-percentage-point margin of error that Fair has set for his model. So the bottom line, based on the long-term economic statistics, is that the election is up for grabs. And that's something the pundits told you 10 days ago.

Update for 8:15 p.m. ET: This week we addressed the presidential platforms on science and technology. To find out where your congressional candidates stand on science and technology issues, check the SHARP Network's database, presented by Scientists and Engineers for America. (SHARP stands for "science, health and related policies.")

Keep up with the presidential race at Politics.msnbc.com.