MIT Department of Electrical Engineering & Computer Science

E E C S

Bargaining Behaviors in Software Agents

David Cliff
MIT, EECS and AI Lab

Monday, October 27, 1997
4:00 PM (refreshments 3:45)
Edgerton Hall, Room 34-101
EECS Colloquium

Abstract

I will talk about developing simple adaptive mechanisms that allow software agents to engage in bargaining behaviors in market-based environments. Groups of agents with such mechanisms could be used in applications including market-based control, internet commerce, and modelling existing or planned economic systems.

The continuous double auction (CDA) is a market structure noted for its speed and efficiency. The CDA is seen in action in the "trading pits" of many international financial derivatives markets: human traders bargain by shouting prices of bids or offers, adjusting the prices depending on activity in the market, and eventually strike deals or drop out.

In 1993, Gode and Sunder demonstrated that human-like CDA market behaviors could be achieved by "zero-intelligence" traders, that randomly generate prices for offers or bids. This implies that the memory or intelligence of agents in the CDA counts for little. I'll present a critique of Gode and Sunder's results, demonstrating failures that can be predicted from analysis of their system, and then go on to discuss some preliminary work in creating simple software agents that behave more like humans in CDA markets.

This talk describes work done while I held a Visiting Academic post at Hewlett-Packard Laboratories Bristol, UK.


URL of this page: http://www-eecs.mit.edu/AY97-98/events/8.html
Created: Oct 16, 1997  | Modified: Oct 16, 1997
This event is from the MIT EECS 1997-98 archive.  | Current events
To MIT EECS home page  | Your comments and inquiries are welcome.