Tuesday, October 18, 2011

Looks like Intraders have discovered Hibbs

On Intrade this morning, Obama's re-election probability is 47.9%, which is the low point for this contract:





This low-ish probability of re-election is now in line with what my main man Doug Hibbs has been predicting via his "Bread & Peace" model, namely Obama loses (Hibbs is predicting 46.2% of the vote for Obama):




(clic the pics for a more Romnian view)

Hibbs has been predicting an Obama loss since May, which is when the Intrade Obama contract peaked at 70% (due to his RIP OBL bounce).

All hail the Stormin' Mormon?

6 comments:

John Thacker said...

Hibbs is predicting a vote share; Intrade is predicting a probability of re-election. Those aren't the same thing.

As a probabilist, I feel bound to complain.

Jeffrey said...

I never comment (only enjoy), and yet I too was compelled to say what John pointed out.

Angus said...

Dear John and Jeffery: I am going to pull a Charles Barkley and say I was mis-quoted.

I didn't mean to give the impression that re-election odds and vote percentages were the same, but I admit the way the post was written very clearly gave that impression.

I have revised it to be less inaccurate. Thanks for reading and commenting!

John Thacker said...

Fair enough.

Certainly agree that <50% of the vote expected is almost always going to be correlated with <50% chance of winning. Not always, but the vast majority of the time.

Dirty Davey said...

That last is of course not true, unless you only consider the two-party vote. Just off the top of my head, three of the last five actual election results have had no candidate with more than 50% of the popular vote, and yet each one had a winner.

John Thacker said...

Yes, I was thinking the two-party vote.

The part about a third party drawing votes is uninteresting and obvious. I think it's more interesting to think of distributions where the expected two-party vote share is below 50%, but the expected probability of winning is above 50%.