Electronic Markets and Multiagent Systems

Trading Agent Competition

The Trading Agent Competition (TAC) (TAC, 2011)) is an annual event that has been held since 2000. The main goal of this event is to encourage high quality research into the trading agent problem. The trading agent problem is a complex decision making problem where autonomous software agents have to negotiate goods to achieve trading goals. For example, a trading agent might have a trading goal to buy components in order to build a computer. Not only does the agent have to try to minimize the purchasing cost but also guarantee that all components are procured (since a computer with a missing component, such as memory or motherboard, is not ready to be assembled).

In the TAC community, the researchers are generally interested in the following research questions (TAC Association, 2011)):

• 1.

  How to design a trading agent? Given an electronic market specification, a trading agent developer normally starts the design process with the identification of key modules that are going to help the agent pursue the trading goals. For example, almost every trading agent has a forecast module where future prices of goods are predicted. This module might help other modules such as a bidding module in order to buy goods and minimize procurement costs.

• 2.

  How to evaluate a trading agent? The evaluation of a trading agent is a key aspect of the development process of a successful agent. The primary motivation for organizing TAC was to create a common electronic market where developers can compare and evaluate their agents.

• 3.

  How do trading agents affect electronic markets? Another important goal of TAC is to evaluate how the competing agents affect an electronic market.

Since 2000, the competition has attracted research groups from around the world to put forth their best efforts at developing automated trading agents for a specific market scenario. The market scenarios in TAC are: