透過您的圖書館登入
IP:18.188.20.56
  • 學位論文

應用多評準決策技術建構最佳化投資組合

Application of Multi-Criteria Decision Making for Optimal Portfolio

指導教授 : 吳泰熙 博士
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


如何在眾多的金融商品中將有限資金做適當的配置,是財務管理領域長久以來不斷被研究的議題,亦即著名的投資組合理論,其發展根基於Markowitz提出的平均數-變異數模型。然而傳統Markowitz的均異模型,是以股價報酬率及報酬率變異數作為績效衡量標準,忽略了投資人在實際操作時亦會考量投資標的基本面資訊。本研究延伸Markowitz的投資組合模型,將投資組合建構過程分為兩階段,第一階段為投資標的選擇,首先利用毛利率、股東權益報酬率與現金流量成長率進行投資候選股初步篩選,將績效表現不佳的個股剔除。接著以12個月績效表現、每股盈餘、股東權益報酬率以及類神經網路預估的股票報酬率作為傳統投資組合報酬衡量指標,以股票下方標準差與季β值做為風險衡量指標,利用資料包絡分析法選取具有相對效率的投資標的。第二階段為資金配置策略,本研究採用三種方法來尋找最適配置比例,以提供投資人作為投資參考依據。此三種方法分別為:(1)資料包絡分析法中,各個受評單位(投資標的)被同儕參考次數佔全部參考次數的比例;(2)多屬性決策模型中,經由TOPSIS、簡單加權、層級加權及ELECTRE等四種方法所求得的投資比例;(3)利用多目標規劃可以同時考量報酬與風險兩種衝突目標的特性,尋找符合投資人預期的妥協解。研究驗證期間為2006年至2009年,期末採買進持有策略,評估在不同資金配置比例下,投資組合之績效表現,並與同期台灣卓越50基金之績效進行比較。 實證結果顯示,加入基本分析的三種投資模式,各季之投資報酬率普遍與台灣卓越50基金之績效不相上下,但每年之平均報酬率皆優於台灣卓越50基金,顯示本研究所建構之投資組合模型,長期而言,有較佳的績效表現。

並列摘要


How to allocate limited capital to numerous financial assets has been studied in the research field of financial management for a long time. It was called portfolio theory, presented by Harry Markowitz who developed the “Mean-Variance Optimization Model”. However, the Markowitz mean-variance model, based on both expected return and variance of return as performance measures, overlooks fundamental financial analyses which investors also consider in their investment process. In this research, we propose a model of portfolio optimization by extending the Markowitz model and adding fundamental analyses as specific objectives related to risk and return. In the present study, the portfolio construction process is divided into two phases. First, we used fundamental analysis and data envelopment analysis as the criteria for initial stock selection. Second, we adopted capital allocation strategy via three methods to find the optimal allocation ratio, including frequency in reference set, multi-attribute decision making model and multi-objective decision making model. We expect to construct an investment system in order to find out the optimal portfolio for investors. The proposed investment procedures have been empirically tested with Taiwan Top50 Tracker Fund from January, 2006 to March, 2010. The result of our analysis shows that, in general, the quarterly performance of our proposed portfolios and Taiwan Top50 Tracker Fund are similar. However, the average annual return rates of our proposed portfolios are better than those of Taiwan Top50 Tracker Fund. We could thus infer that the proposed portfolios in our model have better performance in the long run.

參考文獻


施正宏(2004),結合總體經濟指標及個股財報資料以預測個股漲跌--以台灣電子類股為例,中原大學資訊管理研究所未出版碩士論文。
王義昌(2007),使用基因演算法預測台股加權指數,中原大學資訊管理研究所未出版碩士論文。
吳宗達(2005),台灣共同基金績效評估之分析,元智大學會計研究所未出版碩士論文。
黃光廷(2002),技術分析、基本分析與投資組合避險績效之研究,成功大學會計學研究所未出版碩士論文。
楊伶升(2009),運用類神經網路、投資標的基本面與技術面分析決定最佳投資組合,台北大學企業管理研究所未出版碩士論文。

延伸閱讀