Robust technical trading with fuzzy knowledge-based systems (Forthcoming in “Frontiers in Artificial Intelligence and Application”.)
This paper proposes a framework of robust technical trading with fuzzy knowledge-based systems (KBSs). Particularly, our framework consists of two modules, i.e., (i) a module for preparing candidate investment proposals and (ii) a module for their evaluation to construct a well-performed portfolio. Moreover, our framework effectively utilizes fuzzy KBSs for representation of human expert knowledge: Precisely, in the 1st module, three sets of fuzzy IF-THEN rules implement linguistic technical trading rules, which are designed specifically for getting well performance in different market phases. On the other hand, the 2nd module exploits fuzzy logic to evaluate the prepared investment candidates in terms of multilateral performance measures frequently used in practice. In an out-of-sample numerical experiment, our framework successfully generates a series of portfolios, which show long-term satisfactory records in the prolonged slumping Japanese stock market.