1. Prediction Markets: Tapping the Wisdom of Crowds Yiling Chen Yahoo! Research February 3, 2008

2. NYU Stern 2/3/20082 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. NYU Stern 2/3/20083 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. NYU Stern 2/3/20084 Bet = Credible Opinion  Q: Will Giants win the Super Bowl?  Betting intermediaries  Las Vegas, Wall Street, Betfair, Intrade,... Giants will win the Super Bowl. Info I bet $1000 Patriots will win the Super Bowl. Info

5. NYU Stern 2/3/20085 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. NYU Stern 2/3/20086 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. $1 if Clinton Wins $0 Otherwise

7. NYU Stern 2/3/20087 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. $1 if Clinton Wins $0 Otherwise $1×Percentage of Vote Share That Clinton Wins

8. NYU Stern 2/3/20088 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. $1 if Clinton Wins $0 Otherwise $1×Percentage of Vote Share That Clinton Wins $1 if Patriots win $0 Otherwise

9. NYU Stern 2/3/20089 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. $1 if Clinton Wins $0 Otherwise $1×Percentage of Vote Share That Clinton Wins $1 if Patriots win $0 Otherwise $f(x)

10. NYU Stern 2/3/200810 Prediction Market 1, 2, 3  Turn an uncertain event of interest into a random variable  Hillary Clinton wins election? (Y/N) => 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 $1 if Hillary Clinton wins election $0 otherwise

11. NYU Stern 2/3/200811 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. NYU Stern 2/3/200812 http://intrade.com Screen capture 2008/02/02 Bird Flu Market

13. NYU Stern 2/3/200813 http://intrade.com Screen capture 2008/02/02 Search Engine Market Shares

14. NYU Stern 2/3/200814 Super Bowl Markets

15. NYU Stern 2/3/200815 Function of Markets 1: Get Information  price  expectation of r.v. | all information (in theory, lab experiments, and empirical studies)

16. NYU Stern 2/3/200816 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. NYU Stern 2/3/200817 Function of Markets 1: Get Information  price  expectation of r.v. | all information (in theory, lab experiments, and empirical studies) Value of Contract ? $1 if Patriots win, $0 otherwise

18. NYU Stern 2/3/200818 Function of Markets 1: Get Information  price  expectation of r.v. | all information (in theory, lab experiments, and empirical studies) Value of Contract Payoff Event Outcome $1 $0 Patriots win Patriots lose ? $1 if Patriots win, $0 otherwise

19. NYU Stern 2/3/200819 Function of Markets 1: Get Information  price  expectation of r.v. | all information (in theory, lab experiments, and empirical studies) Value of Contract Payoff Event Outcome P( Patriots win ) 1- P( Patriots win ) $1 $0 Patriots win Patriots lose ? $1 if Patriots win, $0 otherwise

20. NYU Stern 2/3/200820 Function of Markets 1: Get Information  price  expectation of r.v. | all information (in theory, lab experiments, and empirical studies) Value of Contract Payoff Event Outcome $P( Patriots win ) P( Patriots win ) 1- P( Patriots win ) $1 $0 Patriots win Patriots lose $1 if Patriots win, $0 otherwise

21. NYU Stern 2/3/200821 Function of Markets 1: Get Information  price  expectation of r.v. | all information (in theory, lab experiments, and empirical studies) Value of Contract Payoff Event Outcome $P( Patriots win ) P( Patriots win ) 1- P( Patriots win ) $1 $0 Patriots win Patriots lose Equilibrium Price  Value of Contract  P( Patriots Win ) Market Efficiency $1 if Patriots win, $0 otherwise

22. NYU Stern 2/3/200822 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. NYU Stern 2/3/200823 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. NYU Stern 2/3/200824 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. $1 if $0 otherwise

25. NYU Stern 2/3/200825 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. NYU Stern 2/3/200826 Financial Markets vs. Prediction Markets Financial MarketsPrediction Markets PrimarySocial welfare (trade) Hedging risk Information aggregation SecondaryInformation aggregationSocial welfare (trade) Hedging risk

27. NYU Stern 2/3/200827 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, 1999][Oliven 1995][Rietz 1998][Berg 2001][Pennock 2002]  HP market beat sales forecast 6/8 [Plott 2000]  Sports betting markets provide accurate forecasts of game outcomes [Gandar 1998][Thaler 1988][Debnath EC’03][Schmidt 2002]  Market games work [Servan-Schreiber 2004][Pennock 2001]  Laboratory experiments confirm information aggregation [Plott 1982;1988;1997][Forsythe 1990][Chen, EC’01]  Theory: “rational expectations” [Grossman 1981][Lucas 1972]  and more …

28. NYU Stern 2/3/200828 An Incomplete List of Prediction Markets  Real Money  Iowa Electronic Markets (IEM), http://www.biz.uiowa.edu/iem/http://www.biz.uiowa.edu/iem/  TradeSports, http://www.tradesports.comhttp://www.tradesports.com  InTrade, http://www.intrade.comhttp://www.intrade.com  Betfair, http://www.betfair.com/http://www.betfair.com/  Gambling markets? sports betting, horse racetrack …  Play Money  Hollywood Stock Exchange (HXS), http://www.hsx.com/http://www.hsx.com/  NewsFutures, http://www.newsfutures.comhttp://www.newsfutures.com  Yahoo!/O’REILLY Tech Buzz Game, http://buzz.research.yahoo.comhttp://buzz.research.yahoo.com  World Sports Exchange (WSE), http://www.wsex.com/http://www.wsex.com/  Foresight Exchange, http://www.ideosphere.com/http://www.ideosphere.com/  Inkling Markets http://inklingmarkets.com/http://inklingmarkets.com/  Internal Prediction Markets  HP, Google, Microsoft, Eli-Lilly, Corning …

29. NYU Stern 2/3/200829 Contracts and Mechanisms  What is being traded? the “good”  Define:  Random variable  Payoff function  Payoff output  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

30. NYU Stern 2/3/200830 Contracts  Random variables (Questions to ask)  Binary, Discrete  Tomorrow or  Sales revenue $200k  Continuous  interest rate, temperature, vote share …  Clarity  “Clinton wins”  “Saddam out”

31. NYU Stern 2/3/200831 Contracts  Payoff functions  Winner-takes-all (Arrow-Debreu)  Index, continuous  Dividend, pari-mutuel, option: max[0, s-k], arbitrary function  Payoff output  Real money, play money, prize, lottery $1 if $1  vote share

32. NYU Stern 2/3/200832 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 demand=supply  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. NYU Stern 2/3/200833 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. NYU Stern 2/3/200834 Pari-Mutuel Market  E.g. horse racetrack style wagering  Two outcomes: A B  Wagers: AB [Source: Pennock]

35. NYU Stern 2/3/200835 AB Pari-Mutuel Market  E.g. horse racetrack style wagering  Two outcomes: A B  Wagers:  [Source: Pennock]

36. NYU Stern 2/3/200836 AB Pari-Mutuel Market  E.g. horse racetrack style wagering  Two outcomes: A B  Wagers:  [Source: Pennock]

37. NYU Stern 2/3/200837 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 prob’s  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. NYU Stern 2/3/200838 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. NYU Stern 2/3/200839 Iowa Electronic Markets (IEM) http://www.biz.uiowa.edu/iem 2008 U.S. Presidential Democratic Nomination Markets $1 if Hillary Clinton wins $1 if John Edwards wins $1 if Barack Obama wins $1 if “other” wins [source: http://iemweb.biz.uiowa.edu/graphs/graph_DConv08.cfm, as of 2/2/08]http://iemweb.biz.uiowa.edu/graphs/graph_DConv08.cfm

40. NYU Stern 2/3/200840 IEM Winner Takes All Market $1 if Democrat votes > Repub $1 if Republican votes > Dem price=E[R]=Pr(R)=0.415 [Source: http://www.biz.uiowa.edu/iem/, as of 2/2/08] http://www.biz.uiowa.edu/iem/ 2008 US Presidential Election WTA Market

41. NYU Stern 2/3/200841 IEM Vote Share Market $1  vote share of Dem $1  vote share of Repub price=E[VS of Repub]=48.8% [Source: http://www.biz.uiowa.edu/iem/, as of 2/2/08] http://www.biz.uiowa.edu/iem/ 2008 US Presidential Election Vote Share Market

42. NYU Stern 2/3/200842 IEM 1992 [Source: Berg, DARPA Workshop, 2002]

43. NYU Stern 2/3/200843 Example: IEM [Source: Berg, DARPA Workshop, 2002]

44. NYU Stern 2/3/200844 Example: IEM [Source: Berg, DARPA Workshop, 2002]

45. NYU Stern 2/3/200845 Tech Buzz Game  Yahoo!,O’Reilly launched Buzz Game 3/05 @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 http://buzz.research.yahoo.com

46. NYU Stern 2/3/200846 Technology Forecasts  iPod phone  Another Apple unveiling 10/12; iPod Video search buzz price 9/8-9/18: searches for iPod phone soar; early buyers profit 8/29: Apple invites press to “secret” unveiling 8/28: buzz gamers begin bidding up iPod phone 9/7: Apple announces Rokr 9am 10/5

47. NYU Stern 2/3/200847 Tech Buzz Game Performance Based on data from 9/29/05 to 1/27/06, 175 stocks in 44 markets

48. NYU Stern 2/3/200848 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. NYU Stern 2/3/200849 Predicting the CEO  Will Mr. Smith or Ms. Jones be the CEO of company X? $1 if Ms. Jones becomes CEO $1 if Mr. Smith becomes CEO Pr(Mr. Smith) Pr(Ms. Jones) $1

50. NYU Stern 2/3/200850 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 down $1 if Ms. Jones becomes CEO & stock price goes up 1 share of stock, if Mr. Smith becomes CEO 1 share of stock, if Ms. Jones becomes CEO

51. NYU Stern 2/3/200851 CEO Decision Market  Should company X hire Mr. Smith or Ms. Jones as CEO? $1 if Mr. Smith becomes CEO Pr(stock up|Mr. Smith) Pr(Ms. Jones) $1 $1 if Ms. Jones becomes CEO $1 if Mr. Smith becomes CEO & stock price goes up $1 if Ms. Jones becomes CEO & stock price goes up Pr(Mr. Smith) Pr(stock up|Ms. Jones) Which one is higher? Pr(stock up|Mr. Smith)=Pr(Mr. Smith & Stock up)/Pr(Mr. Smith)

52. NYU Stern 2/3/200852 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. NYU Stern 2/3/200853 Decision Markets Give E(O|C) C hoices  FED money policy  Next president  Health care regulation  School vouchers  Who is CEO  Which ad agency O utcomes  GDP per capita  War deaths  Lifespan  School test scores  Stock price  Product sales [Source: Hanson]

54. NYU Stern 2/3/200854 http://intrade.com Screen capture 2008/02/02 Election Decision Markets

55. NYU Stern 2/3/200855 A Combinatorics Example March Madness

56. NYU Stern 2/3/200856 A Combinatorics Example March Madness

57. NYU Stern 2/3/200857 A Combinatorics Example March Madness

58. NYU Stern 2/3/200858 Combinatorics Example March Madness  Typical today Non-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 k Duke > {UNC,NCST} ACC wins 5 games  2 2 64 possible props (implicitly defined)  A bet effects related bets “correctly”; e.g., to enforce logical constraints

59. NYU Stern 2/3/200859 Expressiveness: Getting Information  Things you can say today:  (>43% chance that) Hillary wins  GOP wins Texas  YHOO stock > 30 Feb. 2008  Duke wins NCAA tourney  Things you can’t 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. NYU Stern 2/3/200860 Expressiveness: Processing 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. NYU Stern 2/3/200861 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 & …