Predicting CW

A lot of people seem fascinated with things like the political predictions charts on Intrade. I, frankly, would very much like to be fascinated with them. But in reality, they're deadly boring. Take the betting on the Republican nomination, for example. It's clearly just a kind of lagging indicator of semi-informed conventional wisdom. The crowd didn't have a premonition of Mike Huckabee's rise and -- more damningly -- the crowd didn't have the foreknowledge to recognize that Huckabee's surge was going to lead to a backlash and heightened scrutiny. So he skyrockets up and then down he dips.

On the Democratic side, meanwhile, Hillary is currently valued at 66.9 and I'm absolutely certain that if she loses in Iowa (of which there's certainly more than a 33 percent chance of happening) then she'll plunge.

Not that any of that is wrong, but it's not especially interesting. You'd like to see some kind of noteworthy divergence between what the betting markets are saying and what people in DC speculate about around the water cooler. Instead, you get something that seems to be an equal blend of what people say around the water cooler with national polls. In formal terms, I guess the problem is that everyone is working off the same bits of inadequate information so nothing's really being aggregated. We all know what the latest polls say and that the outcomes in the early states matter more. Put all of us knowing the same thing together and what you get is dull, dull, dull.

Comments

Put all of us knowing the same thing together and what you get is dull, dull, dull.

Isn't the idea that there are some people who know more, e.g. about unreleased polls or looming scandals, and that if they play this should have some effect?

jenny

Posted by jenny | December 28, 2007 10:14 PM

In formal terms, I guess the problem is that everyone is working off the same bits of inadequate information so nothing's really being aggregated.

I'd say it's more an issue of the market actors' errors being nonrandom. These kinds of markets require errors to be random in order to function significantly better that any sound analysis, and, in cases like this, errors are almost certainly driven in significant part by a nonrandom combination of conventional wisdom and partisan wagering. Even in cases of inadequate information, so long as everyone has the same inadequate information available random errors will result in prediction markets offering a sounder interpretation of available information than even pretty good analysis.

Posted by R Johnston | December 28, 2007 10:22 PM

I tried to take advantage of exactly what you said to this to make money but wagering on the likely Iowa winners. Unfortunately it's a pretty serious hassle to try to gamble online nowadays.

Unless of course it's through eTrade...

Posted by DJ | December 28, 2007 10:24 PM

Matt - Isn't "water cooler" a bit hackneyed? Not much talking around the water cooler in reality. Too risky - you may say the wrong thing and pss someone off. Politics is extremely risky to discuss - Usually you can sense political leanings and sensitivities, but it's a fraught area.

Posted by Comment | December 28, 2007 10:38 PM

"On the Democratic side, meanwhile, Hillary is currently valued at 66.9 and I'm absolutely certain that if she loses in Iowa (of which there's certainly more than a 33 percent chance of happening) then she'll plunge."

So how much money are you in for then?

You're saying it's just telling us what we all know and it's wrong?

Posted by The Animal | December 28, 2007 10:50 PM

I love when you blog about things you find uninteresting. It's so interesting.

Posted by brian | December 28, 2007 11:42 PM

Comment asks: "Matt - Isn't "water cooler" a bit hackneyed?"

Thanks for bringing that up. One of the worst aspects of political commentary is how terrible and cliche-ridden the prose is. Just once I'd like to see Matt talk about "chit-chat at the church circle jerk" instead of that frigging water cooler line.

I can dream, can't I?

Posted by MoeLarryAndJesus | December 29, 2007 12:08 AM

MoeLarryAndJesus:

"I'd like to see Matt talk about 'chit-chat at the church circle jerk'"

The water cooler makes me wince, too. I'm usually like, ".... the idle topics we beat out in Huffman code on the chalice with our turgid phalluses."

Matt: as with stock brokers, I'm more impressed when my interlocutor has laid down a substantial portion of lettuce on said prediction.

Posted by Tiny Tim | December 29, 2007 1:03 AM

It's clearly just a kind of lagging indicator of semi-informed conventional wisdom.

Well yes, and you've also described NYSE, NASDAQ and the CBOT.

Put all of us knowing the same thing together and what you get is dull, dull, dull.

Maybe the Atlantic should have a blogger whose central premise is that different people have access to different amounts of data? ;)

Posted by Kolohe | December 29, 2007 2:10 AM

WRT Hillary, while she has a greater than 33% chance of losing Iowa, she has a considerable chance of losing Iowa and still winning the nomination. 2/3 seems about right.

What does seem ridiculous is that Rudy is still the frontrunner for the Republicans. That's ridiculous. It's been clear for some time that Rudy is not terribly likely to be the nominee.

Posted by John | December 29, 2007 2:55 AM

Looters and Moochers like you will never understand, Matt:

The Wisdom of Crowds is the uninformed conventional wisdom of rich people, which is much, much Wiser than the uninformed conventional wisdom of poor people.

Posted by Megan McArgleBargle | December 29, 2007 3:02 AM

Intrade is where all the Peteys hang out. It's the Peteyest.

Posted by antipetey | December 29, 2007 3:03 AM

Don't hate the player, hate the game.

Posted by anti-antipetey | December 29, 2007 3:34 AM

Since interest in Wall St gambling is mainly a Republican and half-Republican (neo-liberal, libertarian, etc) trait, I'd expect Intrade's predictions to be biased that way.

Posted by Gary Sugar | December 29, 2007 6:38 AM

"I'm absolutely certain that if she loses in Iowa (of which there's certainly more than a 33 percent chance of happening) then she'll plunge".

As usual, Matt's tossed off remark is seems more interesting to me than his main topic. I wish I believed this one.

Posted by Gary Sugar | December 29, 2007 6:53 AM

"Not that any of that is wrong, but it's not especially interesting."

Intrade is interesting in two ways:

1) It's an excellent numerical summation of the current CW.

2) It's an excellent way for Petey to make easy money.

While #2 may be of more interest to me, #1 should be of interest to folks like Matthew, who are struggling to come up with words on a primary race they don't have a particularly good grasp of.

Of course Intrade isn't all that predictive. No one with half a brain claims it is. But it's still interesting to see where the CW is at the moment in handy graphs, and it should provide a hook for Matthew to hang his thoughts.

Posted by Petey | December 29, 2007 7:15 AM

"Don't hate the player, hate the game."

Intrade is so fucking easy to beat.

Petey's Current Fair Value:

GOP
40% Romney
30% McCain
15% Thompson
15% Field

Dem
45% Clinton
35% Edwards
20% Obama

So, the obvious plays: going long on Edwards at 6 cents is the best value out there, go short on Giuliani at 30 cents, go long on Thompson at 3.5 cents, and go long on Romney at 22 cents.

Also, if you don't mind tying up your money for a couple of months for a small return, you can earn an 8% return on your investment by going short on Ron Paul with zero risk!

Posted by Petey | December 29, 2007 7:26 AM

I know all sorts of people who were sure that the stock market valued internet stocks way too low because they were going to take over the world in 2000.

It was so easy to make money in the stock market. All you had to do was buy an internet stock.

It was so easy to make money in the housing market. All you needed to do was buy a condo in Miami and rent it out and flip it.

The fact is that Intrade and the Iowa market before it are FANTASTIC predictors of what actually happpens in real life. Politics and business and housing markets.

Obviously you get times where the markets don't work well like the current CDO mess on Wall Street.

Matt:

you are so smart. What odds would you give?
Are you willing to put your money where your mouth is?

Posted by neil wilsom | December 29, 2007 7:34 AM

I'm absolutely certain that if she loses in Iowa (of which there's certainly more than a 33 percent chance of happening) then she'll plunge

"As usual, Matt's tossed off remark is seems more interesting to me than his main topic. I wish I believed this one."

I'm not sure which part of this you don't believe.

You think she'd still be trading above 50 on January 4th if she loses Iowa? (Though she'd still be overvalued at Intrade in that situation, it's highly likely that'd she'd plunge in value compared to today.)

Or you think she's got a better than 2/3rds chance of winning Iowa? (She'd lose the caucuses if they were held today...)

The logic that gets Matthew to his tossed off thought is somewhat nonsensical, but it still gets him to about the right place.

Posted by Petey | December 29, 2007 7:39 AM

"The fact is that Intrade and the Iowa market before it are FANTASTIC predictors of what actually happpens"

You, sir, are dumber than the average bear.

Posted by Petey | December 29, 2007 7:40 AM

while, Hillary is currently valued at 66.9 and I'm absolutely certain that if she loses in Iowa (of which there's certainly more than a 33 percent chance of happening) then she'll plunge.

You are correct. Intrade says there is about a 40% chance of her winning Iowa and a 60% chance of losing.

So what are the odds of her winning the nomination after losing Iowa and what are the odds of her winning the nomination after winning Iowa.

If you assume there is a 95% chance of winning the nomination after winning Iowa then it implies the odds are about 25% to 30% of Hillary still getting the nomination if she loses Iowa.

I don't know about you but that sounds reasonable to me.

Do you disagree?

Posted by Qwerty | December 29, 2007 7:45 AM

"it implies the odds are about 25% to 30% of Hillary still getting the nomination if she loses Iowa."

No. You need to work on your math skills.

To get Clinton up to 67 with your other assumptions, you'd have to assume she'd be a 50/50 shot to still win the nomination after losing Iowa.

Posted by Petey | December 29, 2007 8:00 AM

Yeah, the idea that the markets are predictive should have been shut down after the last Iowa caucuses.

See this graph of the Iowa Electronic Markets.

Petey - I note that Romney has surged in the Petey Markets. I don't disagree, but I'd ask whether you're now underrating Huckabee? He has a serious base of support in a community that could function as a second party establishment, even though he's rejected by the first party establishment. I dunno, but under 15%, given his polling in SC and FL, and his near-certain win in IA, seems wrong to me.

Posted by DivGuy | December 29, 2007 8:47 AM

"The logic that gets Matthew to his tossed off thought is somewhat nonsensical, but it still gets him to about the right place."

Okay, you've watched this market, whereas I'm only measuring my own doubts. If Clinton is smashed in Iowa, ok, she'll plunge. But if she loses Iowa closely, from that in itself, I'd lower the 67 say 10% to 60. But that's just me. I don't play these markets, so I don't know how they typically over- or under-react to this or that.

Posted by Gary Sugar | December 29, 2007 8:57 AM

"I'd ask whether you're now underrating Huckabee? He has a serious base of support in a community that could function as a second party establishment, even though he's rejected by the first party establishment. I dunno, but under 15%, given his polling in SC and FL, and his near-certain win in IA, seems wrong to me."

Best I can tell, Huckabee is unnominatable.

There's a big difference between 1) winning early primaries with 25% - 30% of the vote in a multi-candidate field, and 2) getting up to 50% once the field has been winnowed.

I think Huckabee can accomplish the first, but you need to do the second to actually win the nomination. And assuming Huckabee makes it to the final pairing, I'd bet the farm on the anti-Huckabee candidate.

-----

"I note that Romney has surged in the Petey Markets."

Yup. I didn't think the LDS factor was going to be surmountable for him, but Huckabee's consolidation of the Chrisianist vote actually seems to work out nicely for him.

And for all of Giuliani's bleating about Florida being his firewall, I think the Sunshine State is actually Willard's firewall. With Crist and Jeb, along with Romney's riches, how does he lose Florida?

Posted by Petey | December 29, 2007 9:05 AM

I think Matt's wrong to say that the electronic markets reproduce conventional wisdom. I just (Sat. morning 9:30 EST) went and looked at the very handy Slate page that aggregates and graphs the electronic political markets ("Why Vote When You Can Bet?"). Intrade and Iowa Electronic show Hillary declining and Obama rising until about December 20, but since then Hillary has been rising and Obama falling steadily. The third electronic market (New Futures) shows Hillary and Obama basically steady since about mid-December.

That's not CW. I'd say CW is "Hillary is falling, the Dem race is tightening, and it's too close to call in the early states." The markets are saying something very different. I don't know which is right, but it's certainly not true that the e-markets are just repeating pundit CW.

Posted by Richard Riley | December 29, 2007 9:53 AM

The liquidity is terrible. If there were some real money behind these predictions, Ron Paul wouldn't be trading near 7. If you were certain that Ron Paul wasn't going to win the GOP nomination, you could hit every bid in the order book and then, if/when Paul is eliminated, you'd collect about 375 bucks. Unless intrade allows you to leverage your account capital, you'd have to put up over 6500 bucks to put on that position. If Ron Paul hangs around for 6 months, you'd end up with an annualized return of about 10%. Which sounds great, but because there's not much liquidity there, it's only 375 bucks which makes it uninteresting. On the other hand, if I'm a fringe candiate with some extra money, like Ron Paul, I'd put some bids out on intrade to make my chances look better since some pundits seem to like to glean information from it.

Posted by Mark | December 29, 2007 11:00 AM

But wait, aren't "markets" and "conventional wisdom" supposed to be more or less mirrors of each other?

When conventional wisdom and the markets are at odds, that's usually cause for some kind of alarm. In the stock market, it usually means someone is making a large amount of money at someone else's expense. In the case an intrade divergence from conventional wisdom would be indicative of something, well, unconventional.

Although agreement between convention and the markets is no guarantee of there being nothing untoward happening, it at least is somewhat assuring.

And, yes, it's also a fantastic way to make a stable return on your investment...

Posted by cheem | December 30, 2007 8:53 PM