Friday, October 30, 2009

Prediction Markets Revisited

Those of you who have been receiving this letter for years know that prediction markets are one of my pet subjects. The recipient list has grown a great deal of late, though, so I thought it might be a good time to revisit the whole concept. (Apologies to those of you for whom this is familiar turf.) Prediction markets are a wonderful source of non-biased information, which is something that's increasingly difficult to find.

A prediction market is a real money market where you can bet on just about anything that has a binary outcome, i.e. it will happen or it won't happen. For instance, "A magnitude 9.0 earthquake will occur anywhere on earth before December 31," or "Where the Wild Things Are will win the Oscar for Best Picture." Contracts trade between $0 and $10. At expiration, winning bettors collect $10, while losers forfeit their entire bet. So let's say a contract trades at $4, a bettor who believes the event will occur puts up $4 to make a profit of $6, while an investor who thinks the opposite can short the contract, putting up $6 to make $4. We can directly infer that the "market" believes there to be a 40% probability of the event occurring.

Let use a real example. Currently, "Democrats to win White House in 2012" is trading at 65%, meaning things look pretty good for the Dems, at least right now. If you think the odds are higher than 65%, you simply put up $6.50, and after the election you get your $6.50 back, as well as another $3.50, assuming you were correct. It might seem like this is a long time to wait for a payoff, but these contracts trade continuously, so one can play shorter term trends. Here you can see the "Dems to win White House" contract over time:

(Interestingly, another contract allows one to bet on who the Democrat nominee will be, and Obama's odds are only 82% (is Hillary in the wings?). We can then infer that Obama's odds of being the next president are 53% [0.65 * 0.82].)

Now, what is so cool about all this, exactly? It isn't, at least for me, the appeal of having yet another way to gamble money. Rather, it is because these markets represent great information flow about future events, some of which we care a great deal about. Yes, some are (entertainingly) frivolous, but many are quite important. For instance, what could be more relevant right now than getting a handle on whether health care reform will pass? Yes, there are polls, but prediction markets have proved far more accurate, because they are anonymous and use real money.

Prediction markets are rooted in the "wisdom of crowds" concept (see James Surowiecki's excellent book by the same name), which holds that the aggregated wisdom of a group of random people is frequently more accurate than the smartest single member of the group. Surowiecki's opening example comes from an early 20th country fair in England where fairgoers, for a sixpence, participated in a contest to guess the weight of an ox. The guesses had a high degree of variance, but the average guess was, ultimately, only one pound away from the actual weight. This was considerably more accurate than the guesses of some butchers who were present.

When I was teaching my Yale class, each year I would perform much the same experiment. I would pass around a glass jar filled with change and ask the students to guess the total dollar amount. As with the ox, many individual guesses were comically off, but once again the average was amazingly accurate.

The best site for prediction markets is I just checked, and here is where just a few of the markets are trading:

Event Probability

Healthcare passes (before year end) 10% (wow)

Healthcare passes (before July) 12%

“Cap and Trade” passes (before Dec. ’10) 41%

Romney wins Republican nomination 25%

Pawlenty wins nomination 22%

Palin wins nomination 22%

Palin announces her own TV show (before Dec. ’09) 13%

Republicans recapture the House 30%

Highest marginal tax rate will be > 38% (2010) 21%

The Dow will close above 10,000 this year 54%

The Dow will trade below 6500 this year 16%

U.S. debt subjected to credit warning (before Dec. ’10) 25%

Iran conducts a nuclear test (before Dec. ’10) 16%

Obama sends > 10,000 to Afghanistan 85%

Higgs Boson particle to be observed (Dec. ’10) 11%

Verizon to sell iphone (Dec. ’09) 10%

Freedom Tower to open (Dec. 2013) 45%

BofA repays TARP money (Dec. ’10) 35%

Roman Polanski extradited (Dec. ’10) 70%

There are many, many more. Intrade even lets you suggest contracts. The Roman Polanski one was mine.

Another robust prediction market involves movie grosses, which can be found at This one is slightly different in that no real money is involved (it’s an onshore concern), but nonetheless, it has a record of remarkable accuracy in an area that is notoriously unpredictable. You can see how this would be of great interest to Hollywood insiders. Essentially, bets are made using play money on how much a movie will gross. If a movie "stock” trades at $90, it means that the market believes the movie will gross $90 million.

Here, for instance, we see the movie stock for Avatar, James Cameron’s first film since Titanic. It trades at about $180, and it is interesting to observe its history. There was a long period of hype for this movie that built up expectations to a very high level, and this was reflected in the stock. But then the trailer apparently didn’t live up to the hype, so the stock traded down. $20 says James Cameron checks the price at least once a week.

Another interesting site is, which is an “arena for long term competitive predictions.” Basically, people make very long term bets, with the proceeds going to charity. Bets are hotly debated by posters. Here’s the marquee bet:

Featured Bet - Long Bets

Duration 10 years (02008-02017)

“Over a ten-year period commencing on January 1, 2008, and ending on December 31, 2017, the S & P 500 will outperform a portfolio of funds of hedge funds, when performance is measured on a basis net of fees, costs and expenses.”

Warren Buffett

Protégé Partners, LLC

Stakes: $1,000,000
(Will go to Girls Incorporated of Omaha if Buffett wins,
or Friends of Absolute Return for Kids, Inc if Protégé Partners, LLC wins.

Bring it on! Warren Buffet and a fund-of-funds firm in a $1 million smackdown. So far Buffet is getting his clock cleaned with the S&P down 24% while hedge funds are down 5%.

Entrepreneurs will find many more uses for prediction markets in the future, particularly for use in large corporations as a way to collect uncolored information from employees, suppliers, and customers, who would otherwise have an incentive to shade their projections on such things as sales. This is already happening to a small degree, but I believe it will grow.

The guessing game with the ox was chronicled by a famous British scientist/statistician named Francis Galton, who had wandered into the fair that day in 1905. Galton was a knighted polymath best known for his work in genetics (he was Charles Darwin’s cousin, as it happened) and statistics. Born prominently, Galton was also a noted elitist who coined the term “eugenics.” He believed that certain people were just meant to run things. He asked for the betting slips on the ox because he thought an analysis of the data would illustrate his point, which was that the masses are basically stupid and should stay out of the way. He viewed the results with some dismay. A collection of country rubes smarter than the experts? It wasn’t what he wanted to find, but to his credit he didn't bury his findings.

Which leads me to my last point, and the other reason why I love prediction markets: they are inherently anti-elitist. They are populist by definition. Who doesn’t tire of the pontificators and the pundits, of the nattering nabobs (to borrow from Safire, by way of Agnew)? We are carpet bombed 24/7 by know-it-all philodoxes, and more than telling us what to think, more and more they are telling us what to do. That’s where I get off the bus.

Prediction markets cut through all the b.s.

No comments:

Post a Comment