Election Markets vs. Polls: Super Tuesday Edition

UPDATE 9:12 A.M.
In the end, prediction markets were better predictors of the outcomes in Alabama and Missouri on the Democratic side, while both polls and markets didn't have good early reads on the California victory for Hillary Clinton.

On the Republican side, everyone was surprised by Mike Huckabee's victories in states like Georgia and Tennessee. Chris Masse has the standings here.

The upshot of the Superduper Tuesday is that traders have converted all the new information we got last night into giving both Obama and Clinton a 31 percent chance of being our next president while giving John McCain a 37 percent chance of success in November.

However, this speaks more to the uncertainty over the eventual outcome on the Democratic side. The contract for a Democratic victory in November gives them a 60 percent chance of winning versus a 40 percent chance for the Republicans.

UPDATE 11:23 P.M.
With the exit polls in California showing strong support for Hillary Clinton, her contract in that state shot up 50 points, now giving her a 90 percent chance of winning the state.

UPDATE 10:05 P.M.
Looks like the first discrepancy is settled in favor of prediction markets with Alabama going to Obama.

Relatedly, Simon Jackman points out that there was some serious volatility in Hillary Clinton's presidential nomination contract earlier tonight. Perhaps a leaking of some exit polls?

clinton.wave.gif

ORIGINAL POST 6:20 P.M.

Heading into to the final hours of voting on the East Coast, there are only a few discrepancies between prediction markets and the latest polls. The table below shows the latest polls for each state compared with the predicted outcome as revealed by traders betting on Intrade. I've only listed states for comparison where a poll was conducted since the beginning of the month.

The biggest surprise is Barack Obama's surge in California where the latest polls have showed him running neck-and-neck with Hillary Clinton. But Obama's Intrade California contract has been moving higher throughout the day, at last look giving him a 65 percent chance of winning.

And California is the all-important stopper for Obama, if he can show a good turnout there, his chances of stopping the Clinton nomination march will be greatly strengthened. Via Adam Nagourney:

If Mr. Obama wins California, that is a real momentum blocker for Mrs. Clinton: There are few states in the country that are more identified with the Clinton presidency than this one.

Three notes about the table:
1. Republicans are not included for the simple reason that their race seems to be all but decided in favor McCain.
2. The "Chance Leader Will Win" column uses the polling data to calculate the probability that the leader is actuality ahead. This is a similar measure to the last two columns showing the Intrade probabilities (inspired by Ian Ayres). Incidentally, this method predicts that Massachusetts, New York, Oklahoma and Tennessee are guaranteed majority wins for Clinton while Georgia and Illinois are locks for Obama.
3. States with discrepancies between polls and markets are identified with an asterisk.

State	Poll Date	HC	BO	Chance Leader Will Win	Intrade HC	Intrade BO
Alab.*	4-Feb	45	43	70	20	75
Cal.*	4-Feb	49	45	84	40	65
Conn.	4-Feb	46	48	70	50	51
Dela.	1-Feb	44	42	70	59.5	40
Geo.	4-Feb	29	49	100	26	90
Ill.	3-Feb	30	66	100	1	95
Mass.	3-Feb	56	39	100	85	35
Misso.*	4-Feb	42	45	83	50	50
NJ	4-Feb	46	41	94	85	15
NY	3-Feb	56	38	100	92	10
Okl.	3-Feb	54	27	100	98.8	20
Tenn.	4-Feb	59	37	100	94	8

Add Your Comment

Report Abuse

Archive

Categories

Links