Characterizing Effective Trading Strategies
John Rust, John H. Miller, and Richard Palmer
Key Words:
Double Auction Tournament, Trading Strategy, Learning
This paper presents a comparative analysis of 30 computer trading programs that participated in a double auction tournament held at the Santa Fe Institute in 1990 and 1991. Our objective is to characterize the form of effective trading strategies in double auction markets. We find that a simple rule-of-thumb is a highly effective and robust performer over a wide range of trading environments, significantly out performing more complex algorithms that use statistically-based predictions of future transaction prices, explicit optimizing principles, and sophisticated ‘learning algorithms.’