SI 679 Aggregation and Prediction Markets

Instructor: Rahul Sami (Office hours: Mon 3-4pm 3246E SI-N; Thu 12-1pm 405A West Hall)

1.5 Credit, 7-week course module
First half of Fall 2007 (first class on 4th September)
Tuesday, Thursday 9:00-10:30am, 412WH

Course Goal:

Learn different approaches to aggregating opinions or information from a number of sources in order to come up with a forecast.

Overview:

In many settings, the wealth of information on a particular subject is distributed among many entities, with no single source having all the relevant information. In this course, we will study approaches to elicit and combine this information in order to come up with a forecast or estimate that reflects the combined information of all sources. This idea of aggregating information from multiple sources is an essential ingredient of many applications, including weather forecasting, predicting election outcomes, market research on tastes, and assigning betting odds. Recently, prediction markets have been deployed to aggregate opinions and come up with forecasts on election outcomes, scientific advances, product delivery dates, Academy Award outcomes, and many other events. We will study theoretical and practical aspects of several aggregation tools, including opinion polls, machine-learning techniques to combine or select experts, scoring rules, and prediction markets; we will focus on incentive-centered design techniques to elicit honest and accurate predictions.

Prerequisites

Introduction to game theory: old SI 625, new SI563, or equivalent (may be taken concurrently), or permission of instructor.
Some knowledge basics of probability. If you're not sure you are prepared for this, I have put up a self-quiz here.

Course Schedule

Note: some topics may take a little less or more than one lecture, so this schedule may shift

Lecture 1:
Introduction to information aggregation, with example forecasting applications. Introduction and demo of prediction markets.
Lecture 2:
Group estimation procedures; combining multiple reported probability distributions.
Lecture 3 and 4:
Machine-learning techniques to select the best expert or combination of experts.
Lecture 5:
Incentives to report truthfully. Proper scoring rules.
Lecture 6:
Introduction to Information Markets/ Prediction Markets; Survey of empirical results of effectiveness
Lecture 7:
Theoretical underpinnings: Rational Expectations; learning through market prices.
Lecture 8:
Guest lecture by Yiling Chen (Yahoo!); topic TBA)Lecture 9:
Advanced market forms: Market Scoring Rules, Parimutuel Markets.
Lecture 10:
Other Issues: manipulation, real money vs. play money, privacy and plicy issues
Lecture 11:
Application: Market Research. Virtual concept testing, Yahoo-O'Reilly Buzz Index
Lecture 12:
Application: Predicting Election Outcomes
Lecture 13:
Review and comparison of different approaches.

Course Work and Assessment

3 Assignments 25% Assignments will include problem-solving and short questions based on the techniques studied in class and the readings.
Midterm Exam 25% Midterm exam will test students on understanding and executing the techniques and theory learned, and understanding the strengths and weaknesses of different approaches.
Final Project 50% (due Oct 20th)

Submit a term paper (6-8 pages) on a topic related to prediction markets.