A decision support system for planning stock portfolios based on a two-stage itemset-based approach

Financial Technologies and Intelligent system have become established solutions to support decision making in stock market investments. Hitherto, fundamentals and technical analysis have been milestones for stock price forecasting. Hence they are often taken into account to refine the decision, but it is difficult to assert an analytical combination that is always right for every market or intent. Simultaneously, risk management could avoid torpedos or decisions that do not reflect the risk aversion of the manager, thus specific risk measures are usually consulted to reach robustness and reliability of a financial decision. An effective financial decision support tool should be able to combine the most advanced technological solutions with the experience of financial investors, represented into the knowledge of indicators, ratios and self-judgment in risk aversion. This study aims to implement and analyze what could be a full decision support tool based on the above-mentioned guidelines by a two-stage approach. Starting from an automatic data mining algorithm that extracts stocks considering the historical performances, suitable portfolios are generated and the user knowledge is put into practice by a second step. In particular, a generalization of the well-known Markowitz mean-variance model, that uses whole portfolios proposals rather than single stocks, optimizes through constraints in diversification, risk measures and financial indicators to fulfil different perspectives.