Meghan Danielson
MAT 5900
Research Paper
December 16, 2010
Prediction Markets
Prediction markets are speculative markets created for the purpose of making
predictions and are a tool for collecting group opinion using market principles. They can be
used in many different settings and serve many purposes in society. This paper will examine
the history of prediction markets and the way in which they were put into practice in the United
States. I will give several examples of current prediction markets in America that involve the
use of play money, while also explaining how this concept translates to the betting markets and
exchanges in Europe. With the economic instability that has been occurring in recent years,
Americans want to learn more about their options for financial investments that they can make
on their own.
According to Paul W. Rhode and Koleman S. Strumpf, economics professors at UNC,
prediction markets were correct in forecasting almost all of the 19 presidential elections that
took place between 1868 and 1940. Formal markets existed on Wall Street in the months
leading up to the presidential races. Newspapers reported market conditions to give a sense of
the closeness of the contest in this period as it was prior to widespread polling. The markets
involved thousands of participants and had millions of dollars in volume in current terms. As
already stated, these markets had remarkable predictive accuracy which led the public of the
US to keep participating (“History of Prediction Markets”).
In 1988, The Iowa Electronic Market was created by Robert Forsythe, Forrest Nelson, and
George Neumann. As a part of the IEM, the concept and name of “prediction market” was
officially coined. According to the IEM, the sole purpose of these markets is to provide
forecasts of the outcomes of uncertain events. The markets that make up the IEM are small-
scale, real-money futures markets where contract payoffs depend on economic and political
events such as elections. According to Joyce Berg, one of the IEM governors in 1995 and an
associate professor of accounting at the university, over 3,700 people have open accounts with
total equity of more than $54,000. As elections grow closer, the number of traders always
increases substantially. The incentive behind participating in IEM is that voters don’t have to
tell pollsters the truth. When their own money is on the line, however, people are bound to act
on information they have been presented with and vote much more realistically. Shares on the
IEM can be traded by any person that wants to open an account, which can be opened for as
little as $5 or for as much as $500. Anyone can buy shares and play on the IEM, but most
traders are students in finance and accounting courses who are looking to receive extra
education outside of the classroom (“Iowa Market Takes Stock…”).
Robin Hanson, a professor at George Mason University, was the first to set up and run a
corporate prediction exchange at Xanadu, Inc. in 1990 where employees were able to bet on the
cold fusion controversy, among other company related things. The US Department of Defense
also used prediction markets to publicize a Policy Analysis Market, which speculated
additional topics for markets might include terrorist attacks. The program was quickly
denounced as a “terrorism futures market” and the Pentagon canceled the program. In October
2007, companies from the US, Ireland, Austria, Germany, and Denmark formed the Prediction
Market Industry Association. This association was formed to promote awareness, education,
and validation for prediction markets (“History of Prediction Markets”).
Prediction markets, also known as information or decision markets are speculative
markets created for the purpose of making predictions. While they were originally created for
use in academic settings, as was the case with the IEM, prediction markets are not just for
curious academics anymore. There has been an explosion of interest in the last few years.
People are now given the opportunity to predict everything from how much a new movie will
gross in its first month, to whether or not gasoline prices will rise. Companies, such as Google,
have created internal markets where the voters are the employees (“Prediction Market”).
Google’s prediction markets were launched in April 2005 and are patterned after the
Iowa Electronic Markets. In Google’s terminology, a market asks a question (“How many
users will Gmail have?”) that has 2-5 possible mutually exclusive and completely exhaustive
answers (“fewer than X users”, “between X and Y”, “more than Y”). Each answer corresponds
to a security that is worth a unit of currency, what Google calls a “Gooble”, if the answer turns
out to be correct, and zero otherwise. The trading in Google’s markets is conducted through a
continuous double auction in each security. Short selling is not allowed, but participants can
exchange a Gooble for a complete set of securities and then sell the ones that they choose.
They can also exchange the complete set of securities for currency (Wolfers).
In each quarter, Google creates about 25-30 different markets for its employees to
participate in. The participants receive a fresh endowment of Gooble each time that they can
then turn around and invest in securities. Each of the markets’ questions is designed so that it
can be resolved by the end of the quarter. At the end of each quarter, Goobles are converted
into raffle tickets and prizes are raffled off. The prize budget is $10,000 per quarter, which is
equal to between $25 and $100 per active trader, depending on the number active in a particular
quarter. Participation in the markets is open to active employees as well as some contractors
and vendors. In the first few years, the numbers have averaged out to show that for every 6,425
employees who had a prediction market account, 1,463 placed at least one trade (Wolfers).
Google has worked as a search engine for the past 15 years because it collects
information that is naturally dispersed across the internet. The internal prediction markets at
Google are based on the same principle. Google employees from across the company
contribute knowledge and opinions that are then collected into a forecast by the market. In
addition to making predictions, internal markets can provide insight into how the specific
organization processes information. Prediction markets provide employees with incentives for
truthful revelation and can capture changes in opinion at a much higher frequency than surveys,
allowing one to track how information moves around an organization and how it responds to
external events (Wolfers).
One of the main discussion topics with regards to prediction markets is a question of the
difference between surveys or polls and the prediction markets I keep talking about. Through
my research I was able to find two main differences between these two categories. The first is
that a survey or a poll is a snapshot in time. The question is asked once and the participant
gives their answer. Prediction markets, on the other hand, run continuously over a period of
days, weeks, and months. People have the ability to change their opinion at any time and have
the data of other peoples’ votes available to them which can sometimes cause a person to
change their vote in accordance with the majority. The second main difference is the type of
question. In a prediction market, the question must have an objective and verifiable outcome.
This means that there will never be a simple yes or no question, like “Should we run this
promotion?” Instead, questions will follow more along the lines of “How much additional web
traffic will this promotion generate?” (Inkling).
So why is the term “market” included in the phrase “prediction market”? The first this
to look at is the fact that a prediction market is a way to bring together the opinions of many
different people. These opinions are expressed by buying and selling stocks that represent
possible answers to a question being posed. An example of this is as follows: The question
that is posed to participants is “Who will win the Super Bowl?” For this question, there will be
stock in all of the football teams, so this means a participant can purchase Bears stock, Patriots
stock, 49ers stock, etc. Participants are given virtual cash when they sign up for the market,
and with this cash, they can buy (and sell) shares in these answers to express individual opinion
of what that person thinks will happen. In prediction markets, the stock price represents the
likelihood that the answer is correct. This means that if the Bears stock is at $75, we would say
“there is a 75% chance the Bears will win the Super Bowl,” according to this market (Inkling).
Prediction markets try to “untangle the factors that cause change.” The IEM works a
little bit differently than other markets because it involves the use of small amounts of real
money. In the 2008 election, the IEM let participants bet yes or no on simple questions, such
as “Will Senator John McCain with the election?” These yes or no questions were different
than a survey though because participants were able to go back into their account and switch
answers at any point in time, until the market reached maturity. If the odds in an IEM market
are 50-50, participants could buy a $0.50 contract and receive a $1.00 payout if you win.
Someone on the losing side of the bet would be out $0.50. As bettors reenter the market and
change their opinions, the odds will shift up or down from 50-50. At one point during this
election, the IEM “winner take all” contract for McCain had fallen to just over $0.15. This was
the markets way of saying that he only had 15.5% odds of becoming president (“Iowa Market
Takes Stock…”).
There are many prediction markets in the United States that are open to the public. The
world’s biggest prediction exchange is Betfair, with the parent company having been developed
in August 1999. This market includes bets on sports, in an online casino setting, and online
poker. Intrade is a for-profit company that includes prediction markets from a large variety of
contracts, not including sports. The Hollywood Stock Exchange, Inkling, the simExchange,
Newsfutures, and many others are virtual prediction markets where purchases are made with
virtual money and “stocks” are exchanged with the hopes of winning a prize at the end
(“Prediction Market”).
One of the prediction markets that I found really interesting was bet2give. This is a
prediction market that allows people to use their foresight and predictive powers to raise money
for their favorite cause. If you bet correctly and win, you will effectively be giving the money
that other people bet, to the non-profit organization of your choice. If you bet wrong and lose,
your money will still be going to a good cause, but it will be the good cause chosen by someone
else. With this site, the thrill of betting meets the power of giving. Because no one benefits
from their own participation on the site, bet2give is able to allow users to play with real money
and thus bypass the laws that make gambling illegal in the United States (Bet2Give).
James Surowiecki was a business columnist with The New Yorker. In 2004, he wrote a
book entitled The Wisdom of Crowds, which championed the use of prediction markets. With
this book, Surowiecki had only set out to echo, and in some cases contradict, the ideas of
Charles Mackey’s work Extraordinary Popular Delusions and the Madness of Crowds. His
book ended up showing that, far more often, the crowd gets it right when making a prediction.
The key criteria that he uses to separate wise crowds from irrational ones (as with the crazed
investors in a stock market bubble) are as follows; diversity of opinion, independence,
decentralization, and aggregation. Surowiecki’s book is filled with examples of collective, or
group, intelligence. In his book, Surowiecki describes a case in which a naval officer located a
sunken submarine by asking people with different pieces of specific expertise to take their best
guess at where the submarine was located. Their collective answer ended up within a few
hundred yards of the actual location (Surowiecki).
This idea of scenario forecasts using group intelligence was created by Robert Hanson.
He believed that humans have access to imperfect information and that we tend to follow along
with the “group think” of our peers and those around us. Because not everyone believes the
same things, we often band together in a group with people who believe in our same ideals and
often end up agreeing too much. No single leader can overcome such biases and data gaps to
predict with certainty whether an action will succeed or fail. Hanson suggests that markets can
bridge these gaps because they combine the guesses of many different people, as was explained
in the submarine example (“History of Prediction Markets”).
Prediction markets are often used to predict things such as winners at the Academy
Awards. Every year, Michael J. Mauboussin, an adjunct professor at Columbia Business
School, asks his students to vote on winners in 12 Academy Awards categories (including
obscure ones like film editing and art direction). The article that I read talked about his
findings in the year 2006. The pick the received the most votes from his students, or the
consensus pick, was right in 9 out of the 12 categories. That being said, he noted that of the 47
participating students, only 1 correctly matched the accuracy of the consensus and the average
number of correct answers per ballot was 4.1. This study demonstrates the power of
predictions markets and the idea that, although they are not perfect, they are certainly more
accurate than individual experts of polls. This finding follows the same line of thought as
group intelligence. Mauboussin is quoted making the same comparison; “All of us walk
around with a little information and a substantial error term. And when we aggregate our
results, the errors tend to cancel each other out and what is distilled is pure information.”
(Nocera).
Online gambling is outlawed in the United States through federal laws, as well as many
individual states laws. This causes a problem when dealing with the concept of prediction
markets and online betting. Because of the illegal nature of online gambling in the US, users
operate with “play money” rather than “real money.” This means that the markets are free to
play, participants are given virtual cash to spend at signup, and the best traders are offered
prizes as an incentive to play. Some exceptions to this legality issue are Intrade and
TradeSports because they run operations out of Dublin, Ireland where gambling is legal and
regulated. Therefore, they are out of the jurisdiction of the US. The IEM is also an exception
because it operates from the University of Iowa, under the cover of a no-action letter from the
Commodity Futures Trading Commission that allows bets up to $500. Some prediction
markets also fall into the illegal or controversial category because of the nature of their
markets. A market predicting the death of a world leader might be useful to people whose
activities are related to the leader’s policies, but it is not a wholesome concept. This type of
market is controversial because it could end up turning into an assassination market
(“Prediction Market”).
When a group of people bet on something, the combined intelligence is remarkably
accurate. British scientist Francis Galton first noticed this in 1906. He observed 787 people at
a county fair try to guess the weight of an ox after it was slaughtered. The average of all
guesses was 1,197, which was only one pound less than the correct answer. The IEM is also
extremely accurate in the prediction of US Presidential elections. This market has accurately
foretold each election since 1988 with only 1.33% error rate in voter totals. Prediction markets
are able to work with so much accuracy and such a small error term because they take the
average of all bets being placed. In the example with the slaughtered ox, for every one person
that believes the ox weighs a disproportionately high weight, there will be another person who
under-guesses the weight of the ox. These two will then balance each other out and the final
result will end up being roughly the same as the actual answer (“Prediction Market”).
Because the United States bans online gambling in US based markets, prediction
markets were created to involve betting but with “play” instead of real money. In many
European countries, gambling is legal. Prediction markets have been adapted and applied to
situations involving betting markets. A betting market refers to a betting exchange, which is an
entity that provides “trading” facilities for retail or bookmaker customers to buy and sell
contracts. Betting markets are used extensively in wagers made mostly on horse racing and in
sports markets, but have also expanded to include elections and current market events. In
addition to the use of real money, betting markets and spread betting differ from prediction
market because they are much higher risk and involve real bets that can account for potentially
large gains or losses.
The idea of spread betting was invented by Charles K. McNeil, a mathematics teacher
from Connecticut who became a bookmaker in the 1940s. The idea became popular in the
United Kingdom and was officially put into use in 1974. Stuart Wheeler, a young, unemployed
stock broker in the UK, had the idea to start people trading on gold prices. The idea was to
create an index that would give investor the opportunity to bet on the movement of gold,
without having to actually buy or sell the physical commodity in the market. Wheeler called
this company the Investors Gold Index until the Bank of England objected to trading under this
name and it became known as the IG Index. Although the IG Index was started based on gold
prices, it soon extended its product range to include foreign exchange and commodities
(“Spread Betting Dangers”).
Spread betting is defined as wagering on the outcome of an event. In this kind of
betting, the payoff is based on the accuracy of the wager, rather than on a simple “win or lose”
outcome. The spread is a range of outcomes, and the bet is whether the outcome will be above
or below the spread. The purpose of the spread is to create an active market for both sides of a
binary wager, even if the outcome of an event may appear to be biased to one side or the other.
In every sporting event, there is almost always a favorite and an underdog. In this situation, the
odds are generally too biased towards the favorite when bets are placed on a simple “Who will
win?” wager. The point spread that is created is essentially a handicap towards the underdog.
With spread betting, the question now becomes, “Will the favorite win by more than the point
spread?” (“Spread Betting”)
The bookmaker advertises a spread of 4 in a certain game. If the gambler bets on the
underdog, he will “take the points” and win if the underdog’s score PLUS the spread is greater
than the favorite. This means that if the Underdog scores 8 and the Favorite scores 10, the
gambler wins because 8 + 4 > 10. If the Underdog scores 8 and Favorite scores 13, than the
gambler loses because 8 + 4 5. If the underdog
scores 8 and the Favorite scores 10, the final ratio will be 10 – 4 .
Cowgill, Bo. "Putting Crowd Wisdom to Work." Official Google Blog. Web. 16 Dec. 2010.
.
"Frequently Asked Questions | Inkling." Prediction Markets - Inkling. Web. 16 Dec. 2010.
.
"History of Prediction Markets - Google Search." Google. Web. 16 Dec. 2010.
.
"Iowa Market Takes Stock of Presidential Candidates (Reprinted with Permission of THE WALL
STREET JOURNAL) - Media Kit - Iowa Electronic Markets - The University of Iowa." Henry
B. Tippie College of Business - The University of Iowa. Web. 16 Dec. 2010.
.
Kunz, Ben. "Prediction Markets Meet Wall Street - BusinessWeek." BusinessWeek - Business News,
Stock Market & Financial Advice. Web. 16 Dec. 2010.
.
Nocera, Joseph. "The Future Dividend by the Crowd." The New York Times (2006). Print.
"Prediction Market Definition." Investopedia.com - Your Source For Investing Education. Web. 16
Dec. 2010. .
"Prediction Market." Wikipedia, the Free Encyclopedia. Wikipedia. Web. 16 Dec. 2010.
.
"Prediction Markets: An Uncertain Future | The Economist." The Economist - World News, Politics,
Economics, Business & Finance. Web. 16 Dec. 2010.
.
"Spread Betting Dangers." Financial Spread Betting for a Living. Web. 16 Dec. 2010.
.
"Spread Betting." Wikipedia, the Free Encyclopedia. Wikipedia. Web. 16 Dec. 2010.
.
Surowiecki, James. The Wisdom of Crowds. New York: Anchor, 2005. Print.
"Ten Things to Know about Spread Betting - Times Online." The Times | UK News, World News and
Opinion. Web. 16 Dec. 2010.
.
Wolfers, Justin. "Using Prediction Markets to Track Information Flows: Evidence from Google." Web.
.