Author(s)

N. F. F. Ebecken & B. S. L. P. de Lima

Abstract

This paper focuses on the possibilities of nature inspired methods to the development of trading systems. Those systems were firstly based on neuro-genetic optimizers and indicators, and recently other alternatives have been investigated. Swarm intelligence has been shown to be attractive in different areas and motivated the development of a framework for creating trading signals. This strategy was applied in this work to a specific commodity and the results are presented. Keywords: swarm intelligence, multi-agent systems, trading systems. 1 Introduction Everybody recognizes that there is no universal trading system available, producing a perfect equity curve. Everyone has their own standards, knowledge and specific way of trading. More than that, there are infinite trading systems that do not work. Today we can consider that those trading tools are only aides. In practice for building models (walk-forward), the well known Dr. Bandy’s [1] security selection methodologies recommend one to concentrate attention on some basic points: – Sufficient data to analyze – The price be reasonable – Sufficient liquidity – Typical cycle matches the holding period and drawdown comfort level – Enough profit potential

Keywords

swarm intelligence, multi-agent systems, trading systems.