Structural information in probabilistic framework: A long term portfolio performance analysis

Abstract

Based upon the raw S&P 500 portfolio performance in eight consecutive years, this paper gives an overview analysis of the application of influence diagram to portfolio management. Our findings provide a ground for the question: Is the structure information provide by the expert superior to the one generated by the data? In this study we use an artificial intelligence system called Influence Diagram for portfolio selection. The system was constructed using both expert opinion and an automated system called BN Power Constructor and we found that the expert constructed system outperform human portfolio manager and market in the years of 1995 to 2003.