Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Distributed Prediction Markets modeled by Janyl Jumadinova Raj Dasgupta Weighted Bayesian Graphical Game Outline Introduction Research Problem Janyl Jumadinova Our Solution Raj Dasgupta Simulation Results Conclusion C-MANTIC Research Group Computer Science Department University of Nebraska at Omaha AMEC 2013 1 / 33
Outline Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Problem: Distributed information aggregation - the Janyl Jumadinova Raj Dasgupta interaction among multiple prediction markets. Outline Introduction Solution: A software agent-based distributed prediction Research Problem market model where prediction markets running similar Our Solution events can influence each other. Simulation Results Conclusion Experimental validation: Comparison with other models and trading approaches. 1 / 33
Prediction Market A Prediction market is Distributed Prediction Markets a market-based mechanism used to modeled by Weighted Bayesian Graphical Game - combine the opinions on a future event from different Janyl Jumadinova people and, Raj Dasgupta - forecast the possible outcome of the event based on the Outline aggregated opinion. Introduction Research Problem Prediction markets operate similarly to financial markets. Our Solution Simulation Results Conclusion 2 / 33
Event: 2012 Presidential Election Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Janyl Jumadinova Raj Dasgupta Outline Introduction Research Problem Our Solution Simulation Results Conclusion 3 / 33
Event: 2012 Presidential Election Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Janyl Jumadinova Raj Dasgupta Outline Introduction Research Problem Our Solution Simulation Results Conclusion 4 / 33
Event: 2012 Presidential Election Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Janyl Jumadinova Raj Dasgupta Outline Introduction Research Problem Our Solution Simulation Results Conclusion 5 / 33
Event: 2012 Presidential Election Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Janyl Jumadinova Raj Dasgupta Outline Introduction Research Problem Our Solution Simulation Results Conclusion 6 / 33
Event: 2012 Presidential Election Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Janyl Jumadinova Raj Dasgupta Outline Introduction Research Problem Our Solution Simulation Results Conclusion 7 / 33
Event: 2012 Presidential Election Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Janyl Jumadinova Raj Dasgupta Outline Introduction Research Problem Our Solution Simulation Results Conclusion 8 / 33
Event: 2012 Presidential Election Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Janyl Jumadinova Raj Dasgupta Outline Introduction Research Problem Our Solution Simulation Results Conclusion 9 / 33
Do Prediction Markets Work? Yes, evidence from real markets, laboratory experiments, and theory Distributed Prediction Markets modeled by Weighted Bayesian I.E.M. beat political polls 451/596 [Forsythe 1999, Berg Graphical Game 2001, Pennock 2002] Janyl Jumadinova Raj Dasgupta HP market beat sales forecast 6/8 [Plott 2000] Outline Sports betting markets provide accurate forecasts of Introduction game outcomes [Debnath 2003, Schmidt 2002] Research Problem Market games work [Pennock 2001] Our Solution Laboratory experiments confirm information Simulation Results aggregation [Forsythe 1990, Plott 1997, Chen 2001] Conclusion Theory of Rational Expectations [Lucas 1972, Grossman 1981] and more... 10 / 33
Event: 2012 Presidential Election Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Janyl Jumadinova Raj Dasgupta Outline Introduction Research Problem Our Solution Simulation Results Conclusion 11 / 33
Event: 2012 Presidential Election Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Janyl Jumadinova Raj Dasgupta Outline Introduction Research Problem Our Solution Simulation Results Conclusion 12 / 33
Event: 2012 Presidential Election Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Janyl Jumadinova Raj Dasgupta Outline Introduction Research Problem Our Solution Simulation Results Conclusion 13 / 33
Event: 2012 Presidential Election Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Janyl Jumadinova Raj Dasgupta Outline Introduction Research Problem Our Solution Simulation Results Conclusion 14 / 33
Related Work The trading protocols used in the market: Distributed Prediction Markets - Continuous double auction based protocol [Pennock and Sami, modeled by Weighted Bayesian 2007; Wolfers and Zitzewitz, 2004] Graphical Game - Dynamic Parimutuel Markets [Pennock 2004] Janyl Jumadinova - Automated Market Makers and Scoring Rules [Hanson 2003, Raj Dasgupta 2007; Chen and Pennock, 2007] Outline Trader behavior and trading strategies: Introduction Play money? [Servan-Schreiber et al. 2004] Research Problem Incentive Compatibility and Bluffing [Chen et al., 2009; Jian and Our Solution Sami, 2012; Conitzer, 2009] Risk behavior [Dimitrov et al., 2009; Iyer et al., 2010] Simulation Results Decision making using prediction markets Conclusion Traders can take actions to influence the outcome of the decision/event [Chen and Kash, 2011; Shi et al., 2009; Boutilier 2012] Applications of prediction markets University/classroom use [Ellis and Sami, 2012; Othman and Sandholm, 2010] Biomedical research [Pfeiffer and Almenberg; 2010] 15 / 33
Research Problem Distributed Prediction Markets modeled by Weighted Bayesian Distributed prediction markets Graphical Game - Multiple prediction markets running simultaneously have Janyl Jumadinova Raj Dasgupta similar events. Outline - The expected outcomes of an event in one prediction Introduction market will influence the outcome of a similar event in a Research Problem different prediction market. Our Solution Simulation Results Conclusion Inter-market effects: evidence from financial markets. Inter-market relationship has not been studied in prediction markets. 16 / 33
Our Solution Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game 1 A software agent-based Janyl Jumadinova Raj Dasgupta distributed prediction market model: - comprises of multiple, parallel running prediction Outline markets, Introduction - uses a graphical structure between the market makers of Research Problem the different markets to represent inter-market influence. Our Solution Graphical Games 2 A graphical game-based algorithm that determines the Our Model Our Algorithm best action for the participants in the prediction market Simulation Results using our proposed model. Conclusion 17 / 33
Graphical Games Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Graph representing the local interaction Janyl Jumadinova Raj Dasgupta Provide compact representations Outline Each node represents a player Introduction N i is the neighborhood of player i Research Problem There is an edge ( i, j ) ∀ j ∈ N i Our Solution Graphical Games Our Model |N i | is the degree of local interaction for node i Our Algorithm The maximum k over the graph k = max i |N i | defines Simulation Results Conclusion the complexity of the representation 18 / 33
Graphical Games Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Intuitive graph interpretation : A player’s payoff is only a Janyl Jumadinova function of its neighborhood Raj Dasgupta Implies conditional independence payoff assumption Outline Introduction Research Problem Our Solution Graphical Games Our Model Our Algorithm Simulation Results Conclusion 19 / 33
Model of the Distributed Prediction Market Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Janyl Jumadinova Raj Dasgupta Outline Introduction Research Problem Weighted Bayesian Graphical Games Our Solution Graphical Games Weighted: use weights to model the influence of one Our Model Our Algorithm market maker on others. Simulation Results Bayesian: used to model the uncertainty of one market Conclusion maker about the other market makers and incorporate different types of market makers. Graphical Games: allows to capture the interaction between multiple market makers. 20 / 33
Bayes-Nash Equilibrium Distributed Prediction Markets modeled by Weighted Bayesian Graphical Game Janyl Jumadinova Raj Dasgupta Outline Introduction Research Problem Propose an algorithm to calculate the equilibrium of the Our Solution Graphical Games game efficiently . Our Model Our Algorithm Determines the best action for the market makers in Simulation Results each prediction market using our proposed model. Conclusion The best action gives the market price that returns the maximum utility to each market maker. 21 / 33
Recommend
More recommend
