« Sentence of the Day | Main | Will the informal sector drive third world growth? »
Assorted links
1. How to judge a popular book in 90 seconds or less. It usually takes me less than five.
2. Does InTrade have a Republican bias?
3. The most overpaid actors and actresses?
4. Douglas Holtz-Eakin on taxes: the truth.
5. My colleague Dan Rothschild has a new blog covering Ike and Houston.
6. An awful Op-Ed.
Posted by Tyler Cowen on September 12, 2008 at 10:42 AM in Web/Tech | Permalink
Comments
InTraders have a ponderous free-market bias. I think that correlates us with one major party more than the other.
But take heart: Bush Derangement Syndrome is well represented, at least in the pre-spinoff Tradesports chat pit.
Posted by: caveat bettor at Sep 12, 2008 10:59:57 AM
I am not totally familiar with the mechanics of the various election prediction markets, so maybe there is an obvious answer, but why doesn't arbitrage make their prices converge? I mean, if I can buy a contract in one place for $49 that pays me $100 for an Obama victory, and a contract in another place for $49 that pays me $100 for a McCain victory, I can't lose.
Posted by: y81 at Sep 12, 2008 11:23:06 AM
The Heuristics choosing a book to read seems a little onerous. I've I got just two rules for judging a book by its cover:
1. If the author is on cover, and its not a autobiography/memoir or humor book, the book is going to be crap and isn't worth the time and money to read or even glance at.
2. If they put "Ph.D." or "M.D." after there name, it going to be crap and isn't worth the time and money to read or even glance at.
Posted by: Christopher at Sep 12, 2008 11:25:36 AM
y81: the fee structures at various prediction exchanges 1) become prohibitive for most traders, and 2) present very different incentives for establishing vs. unwinding positions.
And volatile (or consistently low) liquidity is also a factor.
Posted by: caveat bettor at Sep 12, 2008 11:27:49 AM
I have a hard time believing that those who play on InTrade and other political futures markets are actually representative of those who will vote in November.
Where's the data?
Posted by: Richard Sharpe at Sep 12, 2008 11:57:29 AM
Serge Ravitch (aka Adanthar) is a poker pro who believed he could exploit InTrade's weaknesses, and put up about 10K to prove his point.
There was a story in the NY times about it a while back:
http://www.nytimes.com/2008/02/13/business/13leonhardt.html?fta=y
And his progress can be chronicled here:
http://forumserver.twoplustwo.com/41/politics/official-adanthar-tries-crush-market-thread-88375/
I believe he's at least doubled his money. Along the way he's pointed out free money: betting against Obama in the Michigan primary where he wasn't on the ballot. And he got Palin as VP early by finding mention of flight records showing her leaving Alaska.
Posted by: burger flipper at Sep 12, 2008 12:16:54 PM
Maybe caveat bettor and burger flipper have answered my question: if liquidity is low (meaning you can't make big bets) and entrance fees are high, it won't be worth anyone's time to arbitrage between markets. And indeed, an activity that returns $10,000 for several months work would not attract me. (Obviously, I don't know how many actual hours Adanthar spent to make $10,000, but if it was more than 50 to 100, most of us can do better.)
Posted by: y81 at Sep 12, 2008 12:51:56 PM
burger flipper--I certainly find Ravitch's claims to be credible. I deposited $3,000 in Tradesports back in 2005 (prior to the Intrade spinoff), and have withdrawn over $1,000 in profits since that time. This is without any interest paid to my account, nor credit of the $100 wire refund, and with transaction costs between 2-5% of trade value.
Even though I've made consistent profit in Tradesports, as well as trading the recession and global warming contracts on Intrade, I've had my worst run getting long Giuliani last year.
I'm still above water though, which is important because I promised my wife I wouldn't make any further deposits. If you think about it, without interest and with the large fees plus the zero sum outcomes on the futures contracts, almost everyone loses.
The winners are everyone who gets to look at the data without paying a cent. Hopefully, the superiority of prediction markets over polls and experts will lead to better public policy in the future. Then, pretty much everyone wins.
If Ravitch wins--and I think he does--then he is also serving the consumers of Intrade prices, by getting us closer to the truth.
Posted by: caveat bettor at Sep 12, 2008 12:56:36 PM
All of these markets have a pretty large amount of liquidity, so the large difference in price is very bizzare. Robin Hanson suggests that the inability is a sign of insufficient investor incentive to respond to arbitrage opertunity.
This is possible, but I think the problem is likely capital constraints. There have been, approximately, 144 thousand trades on the "McCain wins" contract in the last week, mostly after the convention. If that consists of a huge number of enthusiastic McCain supporters who over-estimate the probability of victory, it would require a huge amount of capital(Well, not much on a global scale. But I doubt Hedge Funds are involved on Intrade) to cause prices to converge to true levels.
It seems that volume has spiked on the other markets, so I think its fairly likely that the number of "Buy Obama on Intrade, Short Obama on [other market]" has been fairly high, but still unable to counteract the wave of Pro-McCain enthusiasm that has hit intrade.
So Democrats should not be too worried about McCain surging ahead on Intrade. Give the markets a week or two to settle down and build capital.
Posted by: David Shor at Sep 12, 2008 12:56:38 PM
David Shor: You may be right. Prediction markets are certainly fallible.
But they are also less fallible than polls. Here is my 2008 state primary experiment between Zobgy and Intrade:
http://caveatbettor.blogspot.com/2008/05/final-intrade-v-zogby-showdown-results.html
Posted by: caveat bettor at Sep 12, 2008 1:03:38 PM
Re: Ike and Houston. Here's a plug for the best weather (hurricane) analysis, I've found. Well written, replete with explanations of all aspects of hurricane dynamics, and forecasts that have proved to be very accurate within the limits of the science and are current ability to forecast multifactorial events like this. Jeff Masters is a University of Michigan meterologist and former Hurrican Hunter team member
Posted by: houstonmike at Sep 12, 2008 1:27:48 PM
Caveot Bettor,
I'm not saying markets are fallible, most of the time they are quite accurate. I study polls a lot at my site(Shameless plug, stochasticdemocracy.blogspot.com), and I'm pretty aware of their failings.
But in the short term, weird events can throw off markets if they are not sufficiently capitalized. The proof that there exist no arbitrage opertunitys in option markets require the presence of "well-capitalized" investors who can take advantage of massive public stupidity.
And I don't think that there are exactly hedgefunds or investment banks on intrade. So it takes some time to gather capital and correct these inbalances.
Posted by: David Shor at Sep 12, 2008 2:26:30 PM
Dave Shor: As a person who has a hedge-fund type job in an investment bank, I'd have to say that hedge funds and investment banks haven't been able to find fair value for, say, a credit derivative, for about a decade now.
Fortunately, I'm on the equities side. Margins are thin, but I haven't blown up (yet).
Great site, btw.
Posted by: caveat bettor at Sep 12, 2008 3:02:03 PM
Tyler, what are your criteria for judging a book in less than 5 seconds?
Posted by: BrianF at Sep 12, 2008 3:10:50 PM
Re: InTrade bias...
We all know that polling data is taken very seriously by those with money on the InTrade exchange. That's evidenced by the plot of state-by-state trading price vs. poll results. [1]
Can someone please explain to me why the distribution in that plot is best fit with a Logistic Curve? [2] (As opposed to say, an Error Function or some other shape?) [3]
[1]: http://electoralmap.net/polling.php
[2]: http://en.wikipedia.org/wiki/Logistic_function
[3]: http://en.wikipedia.org/wiki/Error_function
Posted by: Ed at Sep 12, 2008 4:05:56 PM
I realize its hard to caclulate an individual actors value, but I don't see the significance of the metric they come up with. If you pay an actor 10 million and your movie makes 30 million, that is not as good as paying an actor 20 million and having the movie make 50 million. By the Forbes method, the latter would be much worse.
Posted by: josh at Sep 12, 2008 5:55:09 PM
To Ed;
I can't get a picture of the map, the site is down. But I suspect it actually is fit better by something like an error function.
The voteshare that a candidate will get in a state, with polling average x(poll averaging is a bit complicated, I go over this on my site), can be modeled really well by a normal distribution V~N(x,sigma) if you assume polls follow a random walk.
So probability of victory(and corresponding price of the contract), is going to be P(V>.5)=[Some mess involving the cdf of the normal, which follows an error function].
The error function looks a lot like a logistic curve, so that might be it.
Of course, it could be logistic. But I'd want to do some chi-tests before I say anything.
Posted by: David Shor at Sep 12, 2008 6:26:36 PM
Perhaps the discrepancy between the Iowa Electronic Markets and InTrade is explained by the difference between the rules of the contracts offered at each. As I read them, IEM contracts are based on popular vote share and not the electoral college, while Intrade is about who will actually be elected.
Posted by: Jule at Sep 12, 2008 8:02:17 PM
How very clever of the WSJ author...When you can't find a correlation to imply causation from, you make up a correlation and then imply causation. Aspiring economagicians take note!
Posted by: pants at Sep 12, 2008 10:45:29 PM
That Op-Ed is hilarious
Also, that may be a candidate for most obnoxious blogger, ey?
Posted by: Robert Olson at Sep 12, 2008 11:38:25 PM
I'm actually confused by the attacks toward the op ed ... I don't think he is inventing the correlation at all, or even supporting the correlation he has found. He's merely taking on a ridiculous statement that has found considerable traction amongst the left wing bloggers and pundits, who continually trot out the obnoxious line that the economy or the stock market have done better under democratic presidents. Luskin is just showing that, by that very simplistic reasoning, the opposite could be true. I don't think is trying to say that people should buy into his point as a way of explaining how the economy works, but rather to burst the bubble of the talking point that he originally took on.
Posted by: Jim Rockwell at Sep 13, 2008 1:42:57 AM
I'm actually confused by the attacks toward the op ed ... I don't think he is inventing the correlation at all, or even supporting the correlation he has found. He's merely taking on a ridiculous statement that has found considerable traction amongst the left wing bloggers and pundits, who continually trot out the obnoxious line that the economy or the stock market have done better under democratic presidents. Luskin is just showing that, by that very simplistic reasoning, the opposite could be true. I don't think is trying to say that people should buy into his point as a way of explaining how the economy works, but rather to burst the bubble of the talking point that he originally took on.
Posted by: Jim Rockwell at Sep 13, 2008 1:43:18 AM
The op-ed is correct 90% of the way, undercutting some overly-simplistic views of the (sparse) data in a way that non-economists can get a handle on -- but then draws some too strong conclusions right at the end.
Posted by: John at Sep 13, 2008 3:47:01 AM
I don't think is trying to say that people should buy into his point as a way of explaining how the economy works, but rather to burst the bubble of the talking point that he originally took on.
I don't know about that...I think if he meant the article to be a bubble buster then he would have made it a little more tongue in cheek. besides, Luskin's a McCain adviser.
Posted by: pants at Sep 13, 2008 9:47:07 PM
Wait...so are the people against the op-ed actually advocating the viewpoint that the American economy does better under Democratic presidents BECAUSE they are Democrats (and not Republicans)?
I interpreted the blogger to simply be ridiculing an obvious case of causation/correlation mix-up. Although, from what is quoted, I'm not sure whether the article advocated this stance, and I'm too lazy (and enthralled in my football game) to read it right now. But as for the blogger being on McCain's team, why would he mock the article for trying to switch presidential party affiliations if that were so? The "data" presented a strong case for Democrats, and by mixing presidents, the article made it sound like Republicans were, in fact, better for the economy/market. If our blogger friend supports McCain, would he mock this point?
Also, I interpreted Tyler's labeling the link as an "awful Op-Ed" as referring to the opinion of the blogger towards the WSJ article, not his own towards the blog, but I could be entirely mistaken.
Posted by: Quinton at Sep 13, 2008 10:32:23 PM
