Semi-Daily Journal Archive

The Blogspot archive of the weblog of J. Bradford DeLong, Professor of Economics and Chair of the PEIS major at U.C. Berkeley, a Research Associate of the National Bureau of Economic Research, and former Deputy Assistant Secretary of the U.S. Treasury.

Wednesday, November 08, 2006

Efficient Markets: Hoisted from Comments

Hoisted from comments:

Barkley Rosser:

Brad DeLong's Semi-Daily Journal: Fair and Balanced Almost Every Day: Macaque Says: Vote on November 7!: There is no mispricing (although's predictions may well come out all wrong). So, according to their final predictions based on the bid of the leading candidate, the five most competitive races are MO, VA, RI, MD, and MT. In all the Dem leads narrowly and the Dems need to win all to get their six to get control. Just looking at the median reported bids, the probability that they will do it looks greater than 50%. But, given the dispersions, and most significantly that they all have Dems narrowly leading and must have them all win, it is perfectly consistent that the overall control bet is 2 to 1 in favor of the GOP. It is not 100.

BTW, the Lieberman SecDef switch story is irrelevant to this (although maybe not to the ultimate control outcome) because indeed the story is based on the outcome on election day, prior to any deals with Lieberman. The probabilities coming out of do not look all that ridiculous, although I suspect MT is more likely than RI to go GOP, and based on dispersions, VA is more likely to go GOP than MO, even though the bid spread is a bit higher in VA than in MO. In short, the difference in MO is more statistically significant.

Posted by: Barkley Rosser | November 07, 2006 at 02:49 PM

And Paul Reber:

Following up Barkely & Brad: I was looking at the potential hedged trade of the individual D's in critical Senate races versus the GOP-Holds-Senate contract with great curiousity yesterday afternoon.

Early afternoon, OH, PA and TN were basically assumed to be done (D, D, & R) but there were 4 critical swing races: VA, MO, MT & RI in play. Each was pricing between 60 and 75, suggesting the money put them all at slight D favorites. However the GOP-Holds-Senate was trading at only 70. That seemed very low. BTW, I checked the contract rules and the independents (CT, VT) did count as D for that contract as long as they agreed to caucus with the D's (so it was in play).

If the 4 necessary Senate races were independent probabilities, the GOP-Holds contract was fairly heavily underpriced -- it should have been at least 80 given the other contacts.

As I mulled this curiousity, it occurred to me that the GOP-Holds-Senate contract had implicitly priced in covariance among the individual D-Senate races. That is, the outcomes are not really independent and a "Democratic wave" phenomenon could cause them all to swing together. Which is what appears to have happened. The big win on the hedged trade was if 3 D's and 1 R won (making 4 of the contracts winners -- and you could lose a bit if there was an R sweep, so maybe it's not technically a "hedged" trade?). The actual Dem sweep outcome was close to a wash depending how heavily you weighed the GOP contract. Curiously, when I finally stopped hitting refresh to walk the dog last night, it looked like the 3-1 winning outcome was coming through (Webb was behind in VA) but when I got back, the Dem sweep was apparent. Overall, I'm glad I kept my money in my pocket (it was enough of a rollercoaster to watch the returns come in as it was).

However, I did find watching the Tradesports market evolve over the day very educational. With respect to the apparent pricing in the covariance among races, what makes the market so darn smart? Is it the influence of a few sharp traders who bang the system back to a shrewd equilibrium? Or is there some emergent property of the combined reasoning of the traders that tends to capture the underlying truth? The more I watch the markets, the more impressed I am by them.

Posted by: Paul Reber | November 08, 2006 at 09:55 AM


Post a Comment

Links to this post:

Create a Link

<< Home