| Summary: | 碩士 === 國立高雄大學 === 資訊工程學系碩士班 === 99 === In this thesis I will present a study of a hybrid AI-based methodology for stock selection, which has long been a challenging task in investment and finance. Recent advances in artificial intelligence and soft computing have led to significant opportunities to solve these problems more effectively. Therefore, in this study, the fuzzy theory and support vector machines are employed to rank a set of stocks; and top-ranked stocks are then selected to construct a portfolio. In addition, genetic algorithms were used to optimize the model parameters and perform feature selection simultaneously. Based on several statistical tests, I will show that the portfolios constructed using the proposed method shall outperform the benchmark significantly. The results thus show that the proposed investment approach is effective and robust for stock selection in practice.
|
|---|
