November 1, 2002

12:31 PM

The Political Oddsmaker gives you the odds of a particular candidate winning their election. You can get odds for each of the major U.S. races (Senate, House, and Governors for this election). It claims a 98% success rate in picking the winner since 1996 (more inside.)
posted by pitchblende (15 comments total)

If you select "National" instead of a state, and "All" races, it'll give you the odds for control of each house of congress after the election, and the early odds for The Presidential Election of 2004, including odds on who will be the Democratic nominee. This oddsmaker is by Ron Faucheux, who is editor-in-chief of Campaigns & Elections magazine and teaches at George Washington University.
posted by pitchblende at 12:32 PM on November 1, 2002

Minnesota
Senate
(Latest Update Oct. 30)

Walter Mondale (D) favored over Norm Coleman (R), 20 to 19 (Mondale starts with a 51.3% chance to win)

DEMOCRATIC SEAT AT SERIOUS RISK
posted by languagehat at 1:04 PM on November 1, 2002

I've seen this site before; I still argue how it touts a percentage rate of its predictions being accurate... the site doesn't really flat out say who they think will win, they give betting and percentage odds of who is most likely to win: a subject which is of course invalidated by unknown factors of voter turnout and last-minute campaigning. Even a few days before the election, most of the races are given near-50% odds plus or minus a few decimals in the percentages... in other words, near-even within a margin of error for who wins... this is new?
posted by XQUZYPHYR at 1:06 PM on November 1, 2002

This is great. Now where's the site with a list of bookies who take bets on elections?
posted by son_of_minya at 1:29 PM on November 1, 2002

I'm pretty sure I can get a 98% success rate in picking races for congress as well. Since the incumbent wins 98% of the time , just choose the incumbent. If there's no incumbent you flip a coin. Needless to say, I am fairly unimpressed with their claim.
posted by trust_no_one at 1:35 PM on November 1, 2002

It's easy to get a 98% success rate if you realize that 90-95% of the races are not even close and are not considered competitive.

If they want to impress me, they will have to show what their prediction rate is for competitive races.

Witold
posted by Witold at 1:36 PM on November 1, 2002

A similar idea with a very different way to get the numbers is the Iowa Political Stock Market, where people trade in futures for different electoral things... this year I think all that's up is control of the House and Senate. People buy and sell, prices get set, and you can watch prices vary in response to various events. Prices for Republican-House-Democratic-Senate have been dropping since Wellstone's death, frex.

Way cool. The market that is.
posted by ROU_Xenophobe at 1:41 PM on November 1, 2002

Even a few days before the election, most of the races are given near-50% odds plus or minus a few decimals in the percentages.

I don't think you're looking hard enough. Pull up the House races for a whole state with a mix of Democratic and Republican incumbents. Try Michigan for example, where there is a wide range of odds for the various seats. What's interesting to me is how few of the House races are close, as Witold and trust_no_one point out. And that has to do with not only incumbency, but a smaller, less heterogeneous voting base than a Senate race has, I would imagine.

I'm not all that impressed by his success rate either, but I think this is an interesting tool to see where the close races are leaning.

Also useful are some of his brief explanations for various races. Look up the Maryland governor's race for example. Not super in-depth, but good for a brief overview of the situation.
posted by pitchblende at 1:47 PM on November 1, 2002

Well I'm going with 2% failure rate.

John Cornyn (R) favored over Ron Kirk (D), 8 to 7 (53.3% chance; downgraded from 54.5% chance on Oct. 31; upgraded from 53% on Sept. 19)
posted by thomcatspike at 1:57 PM on November 1, 2002

Oh, poo, mediareport's got all the predict-o-links you'll ever need.
posted by y2karl at 2:03 PM on November 1, 2002

They went out on a limb and gave my Congressman a 50 to 1 chance of winning. Of course, he's running unopposed...
posted by spilon at 2:32 PM on November 1, 2002

Thanks, y2karl. I was actually kind of surprised none of the bloggers I linked to had mentioned the searchable Congressional Quarterly race ratings (see the sidebar on the right). It even includes links to relevant Washington Post articles for each state.

[Btw, I should have MovableType up and running shortly. Yay for joining the real world.]
posted by mediareport at 6:17 PM on November 1, 2002

Weird. Ok, go here and click "About CQ Analysis" to get the searchable rankings page.
posted by mediareport at 6:32 PM on November 1, 2002

Well, Faucheux used to be a politician himself in Louisiana, so obviously he's good at putting the spin on the numbers.

Which I can say he's good at....I should know because in a previous life, I worked for him at C&E.
posted by PeteyStock at 8:31 PM on November 1, 2002

Unfortunately, this year IEM defined their Senate/House control claim poorly. The Republican control claim doesn't pay out if the Republicans control 50 Senate seats, which would lead to actual control by the Republicans. The market is trading at about .75 for non-Republican Senate control, however only in this poorly defined way.
posted by cameldrv at 4:32 PM on November 2, 2002