AAAI-12 Tutorial: Trading Agents

Trading Agents

AAAI-12 tutorial presented by Michael Wellman and Amy Greenwald.

Sunday 22 July 2012, Sheraton Centre Toronto Hotel

*UPDATE* slides now posted here, in four parts

  1. Intro to Trading Agents and Auction Fundamentals (Wellman)
  2. Interdependent Markets (Greenwald)
  3. Complex Market Scenarios (Greenwald)
  4. Continuous Double Auctions, Empirical Game Analysis (Wellman)

Overview

This tutorial frames and motivates the problem of developing automated trading strategies for electronic markets. It surveys the state-of-the-art in analyzing strategies for basic market games, covers examples of more complex (intractable) market scenarios, and presents a general methodology (empirical and game-theoretic) for trading agent design and analysis.

Audience and Goal

Bidding in electronic markets is pervasive throughout the modern economy, from finance (markets in equities, currency, commodities), to electronic commerce (eBay and its ilk), to supply chains (B2B marketplaces), to Internet advertising (search and display ad exchanges). We use illustrative examples of these domains to motivate the problems, which we then address at the more abstract level of generic auction environments, or canonical trading games.

The tutorial is designed for those interested in trading domains as an application area for AI methods, for either research or practice. It will be broadly accessible to an AI audience. Previous exposure to basic concepts of game theory (e.g., the definition of Nash equilibrium) is assumed. Some knowledge of auctions is helpful, but not required.

If successful, tutorial participants will walk away with an understanding of:

  • how autonomous trading agents are or can be applied to real-world markets,
  • what is and is not known about strategies for canonical trading games,
  • key building blocks of successful trading strategies,
  • what to consider in evaluating particular agent designs,
  • how to combine analysis and search to derive new and improved trading strategies.

Tutorial Materials

The tutorial will be largely based on materials from a recent M&C Synthesis Lecture:

MP Wellman, Trading Agents, Morgan & Claypool Publishers, 2010.

It will also draw on:

MP Wellman, A Greenwald, and P Stone, Autonomous Bidding Agents: Lessons from the Trading Agent Competition, MIT Press, 2007.

Schedule of Topics (rough and subject to adjustment)

9:00am

  • What is a trading agent? Definition of automated trading, examples. Program trading in financial markets. Proxy bidding and simple bidding tools for Internet auctions.
  • Auctions and markets. Basic definitions, motivation, general framework.
  • Trading in simple (both operationally and analytically) auctions. Auction-theoretic solutions to sealed-bid and open outcry auctions.
  • Dealing with interdependent values, and the “winner’s curse”.

10:30am   BREAK

11:00am

  • Trading in operationally simple yet analytically intractable auctions.
    • Continuous double auctions (CDAs).
    • Simultaneous ascending auctions.
    • Simultaneous one-shot auctions.
    • Sequential auctions
  • Richer trading domains: Examples and case studies from the Trading Agent Competition
    • dealing with interdependent markets, supply chains, bidding in ad auctions
  • Methodology for trading agent design and analysis.
  • Conclusion. Outlook for trading agents.