Authors:
Paulo André Lima de Castro
1
and
Jaime Simão Sichman
2
Affiliations:
1
Technological Institute of Aeronautics; Intelligent Techniques Laboratory, University of São Paulo, Brazil
;
2
Intelligent Techniques Laboratory, University of São Paulo, Brazil
Keyword(s):
Multiagent systems, automated trading, multiagent architecture.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence and Decision Support Systems
;
Coordination in Multi-Agent Systems
;
Enterprise Information Systems
;
Group Decision Support Systems
;
Market-Spaces: Market Portals, Hubs, Auctions
;
Software Agents and Internet Computing
Abstract:
In order to manage their portfolios in stock markets, i.e., to determine buy and sell signals, human traders use a set of algorithms created by economists, which are based on stock prices series. These algorithms are usually referred as technical analysis. However, traders prefer to use a combination of several algorithms as indicators, rather than choosing a single one. The several signals provided are used to determine the trader order to buy or sell some stocks, or even to decide to not submit any order. In the last years, some researchers have tried to create new algorithms with learning skills in order to produce autonomous automatic traders, some of them using AI techniques. Inspired by the human traders´ decision processes, our architectural approach composes heterogeneous autonomous trader agents in a competitive multiagent system. This architecture allows the use of several algorithms, based on different technical analysis indexes to manage portfolios. We have implemented th
is architecture and we have performed a set of simulation experiments using real-data from NASDAQ stock market. The experimental results were compared to the performance of single agents playing alone, and a better global performance was observed when traders compete with each other for resources. These preliminary results indicate that competition among agents, as proposed here, may reach very good results, even among agents created to act alone in this kind of market.
(More)