Prediction Markets: Tapping the Wisdom of Crowds

1
Prediction Markets Tapping the Wisdom of Crowds

• Yiling Chen
• Yahoo! Research
• February 3, 2008

2
Outline

• Introduction to prediction markets
• What is a prediction market?
• Functions of markets
• Contracts and mechanisms
• Prediction market examples
• Iowa Electronic Markets
• Yahoo! Tech Buzz Game
• Looking forward
• Decision markets
• Combinatorial markets

3
Events of Interest

• Will Giants win the Super Bowl?
• Will Hillary Clinton win the Democratic Primary
  race?
• Will Democratic party win the Presidential
  election?
• Will (Should) Microsoft and Yahoo merge?
• Will US economy go into recession during 2008?
• Will there be a cure for cancer by 2015?
• Will sales value exceed 200k in April?

4
Bet Credible Opinion

• Q Will Giants win the Super Bowl?
• Betting intermediaries
• Las Vegas, Wall Street, Betfair, Intrade,...

5
Prediction Markets

• A prediction market is a financial market that is
  designed for information aggregation and
  prediction.
• Payoffs of the traded item is associated with
  outcomes of future events.

6
Prediction Markets

• A prediction market is a financial market that is
  designed for information aggregation and
  prediction.
• Payoffs of the traded item is associated with
  outcomes of future events.

7
Prediction Markets

• A prediction market is a financial market that is
  designed for information aggregation and
  prediction.
• Payoffs of the traded item is associated with
  outcomes of future events.

1Percentage of Vote Share That Clinton Wins
8
Prediction Markets

• A prediction market is a financial market that is
  designed for information aggregation and
  prediction.
• Payoffs of the traded item is associated with
  outcomes of future events.

1Percentage of Vote Share That Clinton Wins
9
Prediction Markets

• A prediction market is a financial market that is
  designed for information aggregation and
  prediction.
• Payoffs of the traded item is associated with
  outcomes of future events.

1Percentage of Vote Share That Clinton Wins
f(x)
10
Prediction Market 1, 2, 3

• Turn an uncertain event of interest into a random
  variable
• Hillary Clinton wins election? (Y/N) gt 1/0
  random variable.
• Create a financial contract, payoff value of
  the random variable
• Open a market in the financial contract and
  attract traders to wager and speculate

11
Terminology

• Contract, security, contingent claim, stock,
  derivatives (futures, options), bet, gamble,
  wager, lottery
• Key aspect payoff is uncertain
• Prediction markets, information markets, virtual
  stock markets, decision markets, betting markets,
  contingent claim markets
• Historically mixed reputation, but can serve
  important social roles

12
Bird Flu Market
http//intrade.com
Screen capture 2008/02/02
13
Search Engine Market Shares
http//intrade.com
Screen capture 2008/02/02
14
Super Bowl Markets
15
Function of Markets 1 Get Information

• price ? expectation of r.v. all information
• (in theory, lab experiments, and empirical
  studies)

16
Function of Markets 1 Get Information

• price ? expectation of r.v. all information
• (in theory, lab experiments, and empirical
  studies)

1 if Patriots win, 0 otherwise
17
Function of Markets 1 Get Information

• price ? expectation of r.v. all information
• (in theory, lab experiments, and empirical
  studies)

1 if Patriots win, 0 otherwise
Value of Contract
?
18
Function of Markets 1 Get Information

• price ? expectation of r.v. all information
• (in theory, lab experiments, and empirical
  studies)

1 if Patriots win, 0 otherwise
Event Outcome
Value of Contract
Payoff
1
Patriots win
?
0
Patriots lose
19
Function of Markets 1 Get Information

• price ? expectation of r.v. all information
• (in theory, lab experiments, and empirical
  studies)

1 if Patriots win, 0 otherwise
Event Outcome
Value of Contract
Payoff
P( Patriots win )
1
Patriots win
?
1- P( Patriots win )
0
Patriots lose
20
Function of Markets 1 Get Information

• price ? expectation of r.v. all information
• (in theory, lab experiments, and empirical
  studies)

1 if Patriots win, 0 otherwise
Event Outcome
Value of Contract
Payoff
P( Patriots win )
1
Patriots win
P( Patriots win )
1- P( Patriots win )
0
Patriots lose
21
Function of Markets 1 Get Information

• price ? expectation of r.v. all information
• (in theory, lab experiments, and empirical
  studies)

1 if Patriots win, 0 otherwise
Event Outcome
Value of Contract
Payoff
P( Patriots win )
1
Patriots win
P( Patriots win )
1- P( Patriots win )
0
Patriots lose
Equilibrium Price ? Value of Contract ? P(
Patriots Win )
Market Efficiency
22
Non-Market Alternatives vs. Markets

• Opinion poll
• Sampling
• No incentive to be truthful
• Equally weighted information
• Hard to be real-time
• Ask Experts
• Identifying experts can be hard
• Incentives
• Combining opinions can be difficult

• Prediction Markets
• Self-selection
• Monetary incentive and more
• Money-weighted information
• Real-time
• Self-organizing

23
Non-Market Alternatives vs. Markets

• Machine learning/Statistics
• Historical data
• Past and future are related
• Hard to incorporate recent new information

• Prediction Markets
• No need for data
• No assumption on past and future
• Immediately incorporate new information

24
Function of Markets 2 Risk Management

• If is terrible to me,
• I buy a bunch of
• If my house is struck by lightening, I am
  compensated.

25
Risk Management Examples

• Insurance
• I buy car insurance to hedge the risk of accident
  
• Futures
• Farmers sell soybean futures to hedge the risk of
  price drop
• Options
• Investors buy options to hedge the risk of stock
  price changes

26
Financial Markets vs. Prediction Markets
27
Does it work?

• Yes, evidence from real markets, laboratory
  experiments, and theory
• Racetrack odds beat track experts Figlewski
  1979
• Orange Juice futures improve weather forecast
  Roll 1984
• I.E.M. beat political polls 451/596 Forsythe
  1992, 1999Oliven 1995Rietz 1998Berg
  2001Pennock 2002
• HP market beat sales forecast 6/8 Plott 2000
• Sports betting markets provide accurate forecasts
  of game outcomes Gandar 1998Thaler
  1988Debnath EC03Schmidt 2002
• Market games work Servan-Schreiber 2004Pennock
  2001
• Laboratory experiments confirm information
  aggregationPlott 198219881997Forsythe
  1990Chen, EC01
• Theory rational expectations Grossman
  1981Lucas 1972
• and more

28
An Incomplete List of Prediction Markets

• Real Money
• Iowa Electronic Markets (IEM), http//www.biz.uiow
  a.edu/iem/
• TradeSports, http//www.tradesports.com
• InTrade, http//www.intrade.com
• Betfair, http//www.betfair.com/
• Gambling markets? sports betting, horse racetrack
  
• Play Money
• Hollywood Stock Exchange (HXS),
  http//www.hsx.com/
• NewsFutures, http//www.newsfutures.com
• Yahoo!/OREILLY Tech Buzz Game,
  http//buzz.research.yahoo.com
• World Sports Exchange (WSE), http//www.wsex.com/
  
• Foresight Exchange, http//www.ideosphere.com/
• Inkling Markets http//inklingmarkets.com/
• Internal Prediction Markets
• HP, Google, Microsoft, Eli-Lilly, Corning

29
Contracts and Mechanisms

• How is it traded?the mechanism
• Call market
• Continuous double auction
• Continuous double auction w/ market maker
• Pari-mutuel market
• Bookmaker
• Combinatorial
• Automated market maker

• What is being traded?the good
• Define
• Random variable
• Payoff function
• Payoff output

30
Contracts

• Random variables (Questions to ask)
• Binary, Discrete
• Tomorrow or
• Sales revenue lt 100k, 100k - 200k, gt200k
• Continuous
• interest rate, temperature, vote share
• Clarity
• Clinton wins
• Saddam out

31
Contracts

• Payoff functions
• Winner-takes-all (Arrow-Debreu)
• Index, continuous
• Dividend, pari-mutuel, option max0, s-k,
  arbitrary function
• Payoff output
• Real money, play money, prize, lottery

32
Call Market and CDA

• Call market
• Stock market mechanism before 1800
• Orders are collected over a period of time
  collected orders are matched at end of period
• Price is set such that demandsupply
• Continuous double auction (CDA)
• Current stock market mechanism
• Buy and sell orders continuously come in
• As soon as bid ? ask, a transaction occurs
• IEM, TradeSports, NewsFutures

33
CDA with Market Maker

• Same as CDA, but with a market maker
• A market maker is an extremely active, high
  volume trader (often institutionally affiliated)
  who is nearly always willing to buy at some price
  p and sell at some price q p
• Market maker essentially sets prices others take
  it or leave it
• Market maker bears risk, increases liquidity
• HXS, WSE

34
Pari-Mutuel Market

• E.g. horse racetrack style wagering
• Two outcomes A B
• Wagers

Source Pennock
35
Pari-Mutuel Market

• E.g. horse racetrack style wagering
• Two outcomes A B
• Wagers

?
Source Pennock
36
Pari-Mutuel Market

• E.g. horse racetrack style wagering
• Two outcomes A B
• Wagers

?
Source Pennock
37
Bookmaker

• Common in sports betting, e.g. Las Vegas
• Bookmaker is like a market maker in a CDA
• Bookmaker sets money line, or the amount you
  have to risk to win 100 (favorites), or the
  amount you win by risking 100 (underdogs)
• Bookmaker makes adjustments considering amount
  bet on each side /or subjective probs
• Alternative bookmaker sets game line, or
  number of points the favored team has to win the
  game by in order for a bet on the favorite to
  win line is set such that the bet is roughly a
  50/50 proposition

38
Outline

• Introduction to prediction markets
• What is a prediction market?
• Functions of markets
• Contracts and mechanisms
• Prediction market examples
• Iowa Electronic Markets
• Yahoo! Tech Buzz Game
• Looking forward
• Decision markets
• Combinatorial markets

39
Iowa Electronic Markets (IEM)
http//www.biz.uiowa.edu/iem
2008 U.S. Presidential Democratic Nomination
Markets
1 if Hillary Clinton wins
1 if Barack Obama wins
1 if other wins
1 if John Edwards wins
source http//iemweb.biz.uiowa.edu/graphs/graph_
DConv08.cfm, as of 2/2/08
40
IEM Winner Takes All Market
2008 US Presidential Election WTA Market
1 if Democrat votes gt Repub
1 if Republican votes gt Dem
priceERPr(R)0.415
Source http//www.biz.uiowa.edu/iem/, as of
2/2/08
41
IEM Vote Share Market
2008 US Presidential Election Vote Share Market
1 ? vote share of Dem
1 ? vote share of Repub
priceEVS of Repub48.8
Source http//www.biz.uiowa.edu/iem/, as of
2/2/08
42
IEM 1992
Source Berg, DARPA Workshop, 2002
43
Example IEM
Source Berg, DARPA Workshop, 2002
44
Example IEM
Source Berg, DARPA Workshop, 2002
45
Tech Buzz Game
http//buzz.research.yahoo.com

• Yahoo!,OReilly launched Buzz Game 3/05 _at_ETech
• Research testbed for investigating prediction
  markets
• Buy stock in hundreds of technologies
• Earn dividends based on search buzz at Yahoo!
  Search
• Mechanism dynamic pari-mutuel market

46
Technology Forecasts

• iPod phone

• Another Apple unveiling 10/12 iPod Video

price
searchbuzz
47
Tech Buzz Game Performance
Based on data from 9/29/05 to 1/27/06, 175 stocks
in 44 markets
48
Outline

• Introduction to prediction markets
• What is a prediction market?
• Functions of markets
• Contracts and mechanisms
• Prediction market examples
• Iowa Electronic Markets
• Yahoo! Tech Buzz Game
• Looking forward
• Decision markets
• Combinatorial markets

49
Predicting the CEO

• Will Mr. Smith or Ms. Jones be the CEO of company
  X?

1 if Mr. Smith becomes CEO
Pr(Mr. Smith)
1
Pr(Ms. Jones)
1 if Ms. Jones becomes CEO
50
Predicting CEO Outcomes

• How will CEO affect stock prices?
• Alternatively,

1 if Mr. Smith becomes CEO stock price goes up
1 if Mr. Smith becomes CEO stock price goes
down
1 if Ms. Jones becomes CEO stock price goes up
1 if Ms. Jones becomes CEO stock price goes
down
1 share of stock, if Mr. Smith becomes CEO
1 share of stock, if Ms. Jones becomes CEO
51
CEO Decision Market

• Should company X hire Mr. Smith or Ms. Jones as
  CEO?

1 if Mr. Smith becomes CEO stock price goes up
1 if Mr. Smith becomes CEO
Pr(stock upMr. Smith)
Pr(Mr. Smith)
Which one is higher?
1
Pr(Ms. Jones)
1 if Ms. Jones becomes CEO
Pr(stock upMs. Jones)
1 if Ms. Jones becomes CEO stock price goes up
Pr(stock upMr. Smith)Pr(Mr. Smith Stock
up)/Pr(Mr. Smith)
52
CEO Decision Market

• Conditional market
• 1 if stock price goes up and Mr. Smith becomes
  CEO
• 0 if stock price goes down and Mr. Smith
  becomes CEO
• called off if Mr. Smith does not become CEO

1 if stock price goes up Mr. Smith becomes CEO
1 if stock price goes up Mr. Jones becomes CEO
53
Decision Markets Give E(OC)

• Outcomes
• GDP per capita
• War deaths
• Lifespan
• School test scores
• Stock price
• Product sales

• Choices
• FED money policy
• Next president
• Health care regulation
• School vouchers
• Who is CEO
• Which ad agency

Source Hanson
54
Election Decision Markets
http//intrade.com
Screen capture 2008/02/02
55
A Combinatorics ExampleMarch Madness
56
A Combinatorics ExampleMarch Madness
57
A Combinatorics ExampleMarch Madness
58
Combinatorics ExampleMarch Madness

• Typical todayNon-combinatorial
• Team wins Rnd 1
• Team wins Tourney
• A few other props
• Everything explicit(By def, small )
• Every bet independent
• Ignores logical probabilistic relationships

• Combinatorial
• Any property
• Team wins Rnd kDuke gt UNC,NCSTACC wins 5
  games
• 2264 possible props(implicitly defined)
• A bet effects related bets correctlye.g., to
  enforce logical constraints

59
Expressiveness Getting Information

• Things you can say today
• (gt43 chance that) Hillary wins
• GOP wins Texas
• YHOO stock gt 30 Feb. 2008
• Duke wins NCAA tourney
• Things you cant say (very well) today
• Oil down, DOW up, Hillary wins
• Hillary wins election, given that she wins OH
  FL
• YHOO btw 25.8 32.5 Feb. 2008
• No. 1 seed in NCAA tourney win more than No. 2
  seed

60
ExpressivenessProcessing Information

• Independent markets today
• Horse race win, place, show pools
• Stock options at different strike prices
• Every game/proposition in NCAA tourney
• Almost everything Stocks, wagers, intrade, ...
• Information flow (inference) left up to traders
• Better
• Let traders focus on predicting whatever they
  want, however they want.
• Mechanism takes care of logical/probabilistic
  inference

61
Combinatorial Markets

• Single-elimination tournament betting
  (Bracketology)
• Team A wins round 5
• Team A beats team B given that they meet
• Permutation style betting
• Horse A beats horse B
• Horse A finishes at position 1, 2, or 5
• Boolean style betting
• Oil price decreases Democrat wins election
  Number of US Troop in Iraq decreases