Simulating the Madness of Crowds:
Price Bubbles in an Auction-Mediated Robot Market

Ken Steiglitz and Daniel Shapiro
Dept. of Computer Science
Princeton University
Princeton, NJ 08544

VERSION 8/13/96

Abstract

We simulate a multiagent market with production, consumption, and exchange mediated by a sealed-bid double auction. Marked price bubbles and subsequent crashes occur when value-based (fundamentals-driven) and trend-based traders are both present, and the market equilibrium price is ramped up exogenously. Similarly, negative price bubbles and recoveries occur when the equilibrium price is ramped down. Because the simulated market is auction-mediated, we can observe the operations of traders during these events, and study the interactions that produce and resolve bubbles. Some preliminary circuit-breaker experiments are described, in which bubbles are interrupted during their formation.