Prediction Markets Election Contest
>
Over at the NYT's economics blog, Economix, David Leonhardt is running a prediction market contest, looking at odds of various Intrade contests. Pick any 3 of the 20 questions to answer, and the winner gets showered with untold glory and fame.
The contest has an interesting twist: Its based upon the odds at Intrade. That makes the choices less obvious and the strategies for winning the contest more intriguing. David asked me to submit a "guest" contest post for the Economix blog, which I did. You will find it below.
In addition to my post, David graciously invited all of you to participate, saying: "You have a lot of really smart readers. Let's see how well they can do in the contest." Challenge accepted. The contest runs til tomorrow morning at 6am. Go over to the prediction market contest and submit your best guesses.
My approach:
Let me begin with this (weasely) caveat: While I have spent many years studying markets, I am not remotely a political analyst. In fact, I dislike politics, and I especially detest how it has polluted economics.
That said, be aware that my preferences in 2000 were the exact opposite of what actually occurred. My choices were McCain, Bradley, Gore, Bush – and that was pretty much the order they lost in.
And by way of full disclosure, I am probably best described as a Liberal Republican – low taxes, balanced budget, strong defense, no unnecessary overseas involvement, and no government involvement in personal matters (birth control, abortion, gay marriage, etc.) Liberal Republicans are now mythical creatures that no longer exist. I do not recognize the abomination that now calls itself the GOP. I guess that makes me an Independent.
OK, let’s move on to the contest, where whatever mathematical skill I may have could be of some small assistance in this contest.
My first thought: Picking the frontrunners and favorites appears to be a surefire path to the middle of the pack. To “win” this requires identifying which of the long shots really aren’t such long shots after all. In the markets, this is called variant perception – what the crowd (and the money they bet) betting thinks is an unlikely outcome, but is actually a higher probability result than most realize. (The best parallel company example is deep value stocks).
If this were actual money, I would most likely pick two favorites, and a single long shot. Something like:
9. North Carolina: Obama wins (1.9)
3. Obama’s electoral votes 379 or fewer (1.3)
11. Pennsylvania McCain wins (7.1)
[Tiebreaker] Winner popular vote share: 51.5%
That is the fiscally prudent thing to do, but that’s no fun. And since its not real cash, why not go for it?
What might be the variant outcome that most people think is very unlikely? I come up with 3: 1) A McCain victory; 2) A surprisingly tight race; 3) An Obama blowout. A McCain victory is (in my terrible political judgment) highly unlikely. The next variant perception scenario is either a very tight race, or an Obama blowout.
Looking at the questions, there are only a few that fit into this scenario. They are: Q2. Greater than 65%; 3. Obama’s electoral votes (out of 538) 380 or electoral votes more; 7. Obama wins Georgia; 10. Ohio McCain wins; 11. Pennsylvania McCain wins; 12. Virginia McCain wins; 13. Democrat-Republican breakdown 271 or more;
I don’t see McCain winning Ohio, Pennsylvania, or Virginia; He does have a good shot in Florida, but its not a great payoff (2.7 pts). Same with Obama winning Indiana (2.4 pts) and Georgia is a long shot.
That leaves the Obama blowout option the best odds (low possibility) relative to the points (highest points) So the picks I would make are:
2. Greater than 65% turnout;
3. Obama’s electoral votes (out of 538) 380 or electoral votes more
13. Democrat-Republican breakdown 271 or more.
[Tiebreaker] Winner popular vote share: 53.5%
While I give this a less than 20% chance of occurring, it’s the highest payout.
In an election where one of candidates ran many Hail Mary’s, it is only fitting.
One last caveat, so my clients don’t have a heart attack: This sort of low probability, high payoff is the exact opposite of how we run money in the office. It is why we’ve stayed out of trouble most of this year . . .
>
Source:
Prediction Markets and the Election: A Game
David Leonhardt
October 31, 2008, 6:42 pm
http://economix.blogs.nytimes.com/2008/10/31/prediction-markets-and-the-election-a-game/
Monday, November 03, 2008 | 06:30 PM | Permalink
| Comments (0)
add to de.li.cious | 
digg this! | 
add to technorati | 
email this post
Comments
The comments to this entry are closed.
