Matthew Yglesias » Market Panic

Oct 16th, 2008 at 10:02 am

Market Panic

It seems to me that Intrade may need to suspend trading in John McCain shares until some kind of plan can be worked out to recapitalize his campaign.

Comments
24

24 Responses to “Market Panic”

K Says:
October 16th, 2008 at 10:06 am
Finally, we know the real reason the DOW went cliff-diving yesterday – everyone took their money out of the stock market and put it in Obama shares. Obama is
causing the market to tank!!
Ed Says:
October 16th, 2008 at 10:37 am
The McCain campaign is bankrupt….let it go the way of Lehman Bros.
Bob Oso Says:
October 16th, 2008 at 10:49 am
Buy and hold McCain for 2012! Same with Beanie Babies- they’re coming back baby!
Dervin Says:
October 16th, 2008 at 11:29 am
The weird thing is McCain is doing better on the market than he is doing at 538’s projections.

But then again, the market has always been a lagging indicator.
Petey Says:
October 16th, 2008 at 12:19 pm
“The weird thing is McCain is doing better on the market than he is doing at 538’s projections.”

538 is dealing with something different than the Intrade market.

- 538 is always trying to predict what would happen were the election held today.

- Intrade is always trying to find a price for what is going to happen on November 4th.

There are completely understandable situations where 538 could validly be projecting a 100% chance of an Obama win and Intrade could have Obama trading at 90% (or lower) with both 538 and Intrade being perfectly correct.
Rob L Says:
October 16th, 2008 at 12:36 pm
Heh. Anyone catch this piece on dKos back in July?

“Washington Floats Bailout Plan for JohnnieMac”
http://www.dailykos.com/story/2008/7/12/101247/027/690/550518
Glenn Says:
October 16th, 2008 at 12:39 pm
538 is always trying to predict what would happen were the election held today.

That’s incorrect. Nate adjusts his simulations to project the winner on election day, accounting for, for example, the greater error in polls the further out you are. Of course, he can only go by the data as it is known as of the time he runs his simulation, but that’s true for Intrade as well, right?

The one thing I’d say that Intrade can capture that Nate obviously does not is the extent to which Intrade buyers predict some sort of significant event between now and the election that will affect the outcome. And, of course, Intrade can’t lay claim to be a representative sample, nor can it weed out individual desires (as opposed to predictions) as to the outcome.
Petey Says:
October 16th, 2008 at 12:47 pm
“That’s incorrect. Nate adjusts his simulations to project the winner on election day, accounting for, for example, the greater error in polls the further out you are. “

This is gibberish.

When there are infrequent polls to go on (normally further out from the elections) Nate’s non-poll regressions play a larger part in his numbers.

But nothing he’s doing is an attempt to do what you say. And that’s actually a good thing.
Glenn Says:
October 16th, 2008 at 12:51 pm
From Nate’s FAQ, steps 4-6 of his process (emphasis his):

4. Snapshot: Combine the polling data with the regression analysis to produce an electoral snapshot. This is our estimate of what would happen if the election were held today.

5. Projection: Translate the snapshot into a projection of what will happen in November, by allocating out undecided voters and applying a discount to current polling leads based on historical trends.

6. Simulation: Simulate our results 10,000 times based on the results of the projection to account for the uncertainty in our estimates. The end result is a robust probabilistic assessment of what will happen in each state as well as in the nation as a whole.
Glenn Says:
October 16th, 2008 at 12:52 pm
Oh, and Petey? Go fuck yourself.
Petey Says:
October 16th, 2008 at 1:09 pm
“Projection: Translate the snapshot into a projection of what will happen in November, by allocating out undecided voters and applying a discount to current polling leads based on historical trends.”

“applying a discount to current polling leads based on historical trends” tends to become close to irrelevant at this point of the game in heavily polled states.

For example, in Florida at this point, Nate’s regression (aka historical trends) is accounting for only about 2 or 3 percent of his overall data on Florida. His Florida numbers are almost entirely a combination of past Florida state polls and past national polls. In other words, his Florida projections numbers are an attempt to tell what would happen in a Florida election held 3 or 4 days ago.

The regression numbers were far more important in June. At this point, in states with frequent polling, they are very close to being irrelevant. (The exception are infrequently polled states like Arkansas, where national trackers and Nate’s regressions model become important. But those states really are the exception. And even in those exceptions, the numbers are still trying to project an election held a few days ago, not an election held in the future.)

As far as the allocating undecideds, that is dependent on the polling numbers, and also isn’t really a predictor.

“Oh, and Petey? Go fuck yourself.”

Go fuck yourself with a bent rusty nail, Glenn. And when you’re bandaged up, try to figure out what information Nate’s (very excellent) model is actually conveying.
K Says:
October 16th, 2008 at 1:18 pm
“…applying a discount to current polling leads based on historical trends.”

“Historical trends” include the likelihood that the race will tighten as Election Day gets closer. Hence the reason that Nate projects a 7.7 point Obama win if the election were today, but a 6 point Obama win on Nov. 4. The regression is something entirely separate.

In other words, Glenn is right and Petey’s wrong.
Scott de B. Says:
October 16th, 2008 at 1:20 pm
For example, in Florida at this point, Nate’s regression (aka historical trends) is accounting for only about 2 or 3 percent of his overall data on Florida. His Florida numbers are almost entirely a combination of past Florida state polls and past national polls

That’s because, this close to the election, those polls are a very good predictor of what will happen on Nov. 4.
Petey Says:
October 16th, 2008 at 1:36 pm
“Historical trends” include the likelihood that the race will tighten as Election Day gets closer. Hence the reason that Nate projects a 7.7 point Obama win if the election were today, but a 6 point Obama win on Nov. 4.”

Point taken.

This is entirely true, and is indeed an example of how Nate is attempting to project forward in certain ways. But it’s a relatively minor part of the overall data (well less than 10%), and the fact remains that the overwhelming bulk of Nate’s numbers are describing an election held 4 days ago, not an election held three weeks forward.

(And that’s not a bad thing.)

Tangentially, much like the regressions, the “tightening assumption” was a much more important contributor to the data in June than it is now.

“In other words, Glenn is right and Petey’s wrong.”

If this were true, then Nate would be an idiot for having “projected” McCain as the winner in September. But Nate’s not an idiot, because that’s not what he was doing.

His model has always been primarily dependent on the current state of the race, with minor tweaks on the edges where it makes sense.
Adam Says:
October 16th, 2008 at 1:38 pm
No, Petey, you’re once again wrong. Nate’s current projection in Florida (if the election were today) is O+6.1. The projection in Florida for actual election day is O+4.6. That’s a rather large adjustment. You’re confusing the trend adjustment with his regression, which is in fact 2-3% of the total weight in most states, and is used most primarily to correctly judge states like South Dakota that receive no polling. The regression receives the weight of an average poll, whereas the trend adjustment adjusts all polls towards the underdog by 1.5-2 points. Intrade and 538 should in fact match up in a perfect world. The traders just don’t realize that an 8 point lead with three weeks to go and no more debates is virtually insurmountable (though they’re beginning to).
Petey Says:
October 16th, 2008 at 1:40 pm
“That’s because, this close to the election, those polls are a very good predictor of what will happen on Nov. 4.”

It has nothing to do with the proximity of the election. It only has to do with the frequency of polling.

The FL regressions would have been just as ignored in June as they are today had there been more frequent FL polling in June.
Petey Says:
October 16th, 2008 at 1:56 pm
“No, Petey, you’re once again wrong. Nate’s current projection in Florida (if the election were today) is O+6.1. The projection in Florida for actual election day is O+4.6. That’s a rather large adjustment.”

But it’s not just the “tightening factor”. It’s also the “50-50 undecideds allocation factor”.

It a blunt force instrument that works to lessen Obama’s margin in any state where he leads.

“The traders just don’t realize that an 8 point lead with three weeks to go and no more debates is virtually insurmountable”

I don’t think you realize the meaning of “virtually insurmountable”. The qualifier is correctly put there. And I don’t think you understand the power of odds that reach 10 to 1 or 20 to 1.

Traders can quite rationally (and profitably) bet against “virtually insurmountable” leads at the proper odds. I bet on McCain to win the GOP nomination in the summer of ‘07 when he was thought to be “virtually dead” at 20 to 1 and laughed all the way to the bank. I thought about betting HRC to win the NH primary at 20 to 1 when Obama had a “virtually insurmountable” lead, and wish I had acted on that thought.

Obama is close to a lock to win this election, and would be a 95%+ lock to win today, but that doesn’t mean that there isn’t a price considerably below Nate’s 95.1 where McCain isn’t the better bet. I’d say fair value on an Obama Intrade contract right now is somewhere around 90, give or take a couple of points. (And the difference between 90 and 95 isn’t slight – the difference is double.)

No one ever expects the Spanish Inquisition, but you have to factor that chance into a contract’s price. Things can change, even when you don’t see the cause coming even though you’re looking hard.
viagra Says:
March 1st, 2009 at 10:23 am
viagra
Very interesting site. Hope it will always be alive!
tramadol Says:
March 22nd, 2009 at 7:32 am
tramadol
I want to say – thank you for this!
cheap viagra Says:
April 8th, 2009 at 4:11 am
I bookmarked this site. Thank you for good job!
viagra

Matt Yglesias

Market Panic

24 Responses to “Market Panic”

Blogroll