Exploiting stock data: a survey of state of the art computational techniques aimed at producing beliefs regarding investment portfolios

Abstract

Selecting an investment portfolio has inspired several models aimed at optimising the set of securities which an investor may select according to a number of specific decision criteria such as risk, expected return and planning horizon. The classical approach has been developed for supporting the two stages of portfolio selection and is supported by disciplines such as econometrics, technical analysis and corporative finance. However, with the emerging field of computational finance, new and interesting techniques have arisen in line with the need for the automatic processing of vast volumes of information. This paper surveys such new techniques which belong to the body of knowledge concerning computing and systems engineering, focusing on techniques particularly aimed at producing beliefs regarding investment portfolios.