資訊管理學報
魏巧宜;馬麗菁;
頁: 69-95
日期: 2017/01
摘要: 隨著經濟發展，大眾對於投資理財的需求與日俱增，如何協助不同投資時間長短偏好的投資者，找出較佳短、中及長期績效的基金，是一項受關注的議題。本研究以自組織映射圖結合倒傳遞類神經網路及基因演算法，協助不同投資時間長短偏好的投資者，找出較佳短期、中期、長期績效的基金。此外，過去的研究大多是分析單一期間的基金績效，本研究進一步以自組織映射圖結合案例式推理法及基因演算法，進行多年期群集軌跡預測。研究結果顯示無論短、中及長期績效預測結果，皆是以本研究所提出的混合模式預測結果最好。在多年期軌跡的預測分析方面，本研究提出的多年期群集預測方法，在風險趨勢預測效果上，亦優於單一期間預測分析結果。
關鍵字: 混合模式;自組織映射圖;案例式推理;多年期軌跡;基金績效;
Abstract: Purpose- As economic growth continues, the needs for investments are increasing. How to assist investors with various preferences in finding out mutual funds with better performance is an important issue. This study aims to propose a framework to analyze the relationships between fund attributes and performance for different investment horizon and to predict multi-year trajectory. Design/methodology/approach - This study combines self-organizing map, genetic algorithms and back-propagation neural network to analyze the relationships between fund attributes and performance for short, middle and long time horizon. Moreover, this study incorporates the concept of self-organizing map, case-based reasoning and genetic algorithms to conduct multi-year trajectory analysis. Findings- The results show that the proposed hybrid approach yields the best prediction performance for all investment horizons. In addition, the proposed multi-year trajectory analysis is better than single-period analysis in predicting risk trends. Research limitations/implications- Because data period adopted in the empirical study is only five years, long-term stability of the proposed framework has not been verified. Practical implications- This paper provides several implications for investors, fund managers and researchers. Investors can get more knowledge about the relationships between fund attributes and performance for different investment horizon. Fund managers and researchers can pay more attention to multi-year trajectory analysis. Originality/value- This study proposes an approach to analyze the relationships between fund attributes and performance for different investment horizon. Rather than analyzing single-period performance only, this study incorporates several methods in business intelligence to conduct multi-year trajectory analysis.
Keywords: hybrid model;self-organizing map;case-based reasoning;multi-year trajectory;fund performance;
瀏覽次數: 13049 下載次數: 133
引用 導入Endnote
頁: 69-95
日期: 2017/01
摘要: 隨著經濟發展，大眾對於投資理財的需求與日俱增，如何協助不同投資時間長短偏好的投資者，找出較佳短、中及長期績效的基金，是一項受關注的議題。本研究以自組織映射圖結合倒傳遞類神經網路及基因演算法，協助不同投資時間長短偏好的投資者，找出較佳短期、中期、長期績效的基金。此外，過去的研究大多是分析單一期間的基金績效，本研究進一步以自組織映射圖結合案例式推理法及基因演算法，進行多年期群集軌跡預測。研究結果顯示無論短、中及長期績效預測結果，皆是以本研究所提出的混合模式預測結果最好。在多年期軌跡的預測分析方面，本研究提出的多年期群集預測方法，在風險趨勢預測效果上，亦優於單一期間預測分析結果。
關鍵字: 混合模式;自組織映射圖;案例式推理;多年期軌跡;基金績效;
Using Hybrid Models to Predict Domestic Equity Fund Performance and Multi-Year Trajectory
Abstract: Purpose- As economic growth continues, the needs for investments are increasing. How to assist investors with various preferences in finding out mutual funds with better performance is an important issue. This study aims to propose a framework to analyze the relationships between fund attributes and performance for different investment horizon and to predict multi-year trajectory. Design/methodology/approach - This study combines self-organizing map, genetic algorithms and back-propagation neural network to analyze the relationships between fund attributes and performance for short, middle and long time horizon. Moreover, this study incorporates the concept of self-organizing map, case-based reasoning and genetic algorithms to conduct multi-year trajectory analysis. Findings- The results show that the proposed hybrid approach yields the best prediction performance for all investment horizons. In addition, the proposed multi-year trajectory analysis is better than single-period analysis in predicting risk trends. Research limitations/implications- Because data period adopted in the empirical study is only five years, long-term stability of the proposed framework has not been verified. Practical implications- This paper provides several implications for investors, fund managers and researchers. Investors can get more knowledge about the relationships between fund attributes and performance for different investment horizon. Fund managers and researchers can pay more attention to multi-year trajectory analysis. Originality/value- This study proposes an approach to analyze the relationships between fund attributes and performance for different investment horizon. Rather than analyzing single-period performance only, this study incorporates several methods in business intelligence to conduct multi-year trajectory analysis.
Keywords: hybrid model;self-organizing map;case-based reasoning;multi-year trajectory;fund performance;
瀏覽次數: 13049 下載次數: 133
引用 導入Endnote