Has Intrade ever been wrong in a political prediction?

[Happyhippychick]

They have the repubes at 86% for controlling the House, Dems at 60% for the Senate. I'm wondering if they have been wrong, anyone know?

[Tony_FLADEM]

1. They predicted John Kerry would defeat George W. Bush back in 2004

[DireStrike]

3. So... they haven't been wrong yet?

[The Northerner]

6. I thought that the Intrade members predicted that Bush would win in 2004

According to the video below, watch starting at the 2:06 mark (yes I know it features Stossel but it seems worthwhile and you don't have to watch it all):


http://www.youtube.com/watch?v=4Fkv1MAkZEw

[Cali_Democrat]

2. There are still 2 weeks left. n/t

[bemildred]

4. "Past performance is no guarantee of future results." nt

[Dawson Leery]

9. +1

[bemildred]

12. I knew somebody would get it. nt

[The Northerner]

5. They were wrong in 2008 when they predicted that Missouri's electoral votes would go to Obama

but they were nearly accurate on most everything else during the 2008 elections.

I remember that they accurately predicted in 2006 that the House & Senate would return to Democratic control.

[Schema Thing]

7. you'd have to compare the same time frames.

Intrade may look very different in 10 days.

[The Northerner]

11. Indeed

I'd recommend checking back a few days before November 2.

[PoliticAverse]

8. Check also the University Of Iowa's Political markets.

At: http://iemweb.biz.uiowa.edu/quotes/Congress10_Quotes.html

[Awsi Dooger]

10. Depends on the time frame and the percentage

Two weeks out, an 86 is very likely to hold up. That's more than a 6/1 favorite. A 60 is only 3/2.

If you want a reference point, a 60 is the money line equivalent of a 2.5 point favorite in a football game, which is less than home field advantage. An 86 is equal to a 13 point favorite.

In other words, a 60 could easily level out or flip to the same number in reverse without much of a shove. The numbers in that region are very fluid. An 86 probably won't deviate much for a while. There needs to be some play in the number, a bit of uncertainty. Once you get closer to game time, even if the polls don't move at all, that 86 will jump into the low 90s, then mid 90s once the early returns show up. That's the way Intrade works, and probability works. Fewer days/hours remaining for an unforeseen event slowly solidify the lopsided favorites.

I'm a gambler. Intrade made political betting more difficult, there's no question about it. In the '90s and early in the '00s you'd have individual websites putting up their own numbers. That was the wisdom of an individual, the oddsmaker, not a crowd. Gaffes galore. My friends and I were happy to take advantage.

That's extremely rare now. The few sites that put up fixed odds are well aware of Intrade and more or less mirror their prices.

On political races during the fall, Intrade is driven by polls. It's not much different than Nate Silver's work, a mathematical compilation more than insight. Occasionally an oddball late poll will shift conventional wisdom and create a bargain. My rare bets on Intrade these days are when a state with a horrendous polling history errs in the same direction as typical, and bettors don't seem to grasp that they are being misled again. I've never lost a wager on Alaska or Georgia politics. Inevitably the polling is far too friendly to the Democrat and undervalues the most conservative candidate. I wish there were an example in the other direction, a state with overwhelming bargains on a Democrat. I'm not aware of one. New Jersey is known is understate Democratic strength early in a cycle but the numbers on Intrade seemingly take that into account.

In 2004 Intrade had Bush the favorite throughout, until election afternoon. The early exit polls created a stampede of Kerry money and pushed him to 2/1 favoritism for a while. My friend and I got caught up in it and lost a chunk, giving back most of our profit from that cycle.

[Nye Bevan]

13. Intrade does not make political predictions, so they cannot be "wrong".

All they do is make a market. If a sports bookmaker quotes 3/1 odds on the Yankees winning the World Series, and the Yankees go on to win, that does not mean that the bookie was "wrong".

[Statistical]

14. Intrade is rarely wrong right before an event but we are two weeks out.

Meaning it is possible (not likely but possible) that you see that 86% decline to <49% right before election day making intrade "right".

The further from an event the more volatility in the numbers.

If it is still 80%+ 2 days before election day I would trust that a lot more than 80% 2 weeks or 2 months before election day.