整合多屬性決策及模糊分群方法於台灣共同基金之選擇

Abstract

隨著國人理財觀念提升，專業理財之重要性越來越為投資人所重視，因此專業投資已形成一股趨勢，其中以共同基金最受到投資人的青睞。面對種類繁多的共同基金，投資人在選擇基金標的時不免感到困難，而不同投資人對於風險及報酬之偏好程度及心中權重各不相同，故適合不同類型投資人之基金應不相同。因此本研究建構出一個以市場上投資專家及投資人角度為出發點的基金評選模式，以投資人最基本可取得，同時最為在意的基金歷史報酬及風險指標值對基金進行分群及排名，此外因考慮到不同投資人對於報酬及風險偏好的不同，本研究應用多準則評估之決策理論對於報酬及偏好求取各基金指標值的主觀權重，應用至基金分類及排名上，可提供更為客製化的建議。 本研究以台灣共同基金之實際資料進行分析，採用的基金指標值為報酬率（Return）、超額報酬率（α）、標準差（σ）及市場風險（β）。本模式建構出的基金評選模式為首先利用專家意見及DEMATEL演算法確立選定的基金指標之因果關係後，以ANP演算法為權重獲取模型之架構，先就市場上投資專家建議，利用Weighted FCM分群演算法，以基金報酬面及風險面之歷史表現將基金分群，提供投資人大方向的建議。而後利用多屬性決策中的TOPSIS演算法對於各基金進行排名，提供更細節的基金選擇參考。此基金評選模式對於投資人而言十分簡潔易懂，只要瞭解本身是屬於何類型投資人，即可迅速得到基金選擇之建議，另外也可計算本身對於報酬及風險的偏好程度，以權重表示，可進行個人化的基金選擇建議。

As the increase of people’s notion to financial management, the importance of professional financial management is getting investors’ attention. Professional investment has become a trend in which mutual fund is most valued by investors. Facing various kinds of mutual fund, it is common that investors feel difficult when choosing fund target. Besides, different investors have different preferences and weights to risks and profits. As a result, funds which suit different type of investor may vary. This study constructs a fund evaluation model based on the viewpoints of professional investors in the market. The study uses the most easily get and most concerned information by investors, including historical fund return and risk index, to classify and rank funds. Besides, considering different investor has different preference to risks and returns, the research uses multiple criteria decision making theory to calculate subjective weight of each fund index, applying to fund clustering and ranking to provide more customized suggestions. This study uses real data from Taiwan mutual fund, applying fund indexes such as return, excess return (α), standard deviation (σ), and market risk (β). The cause effect relationship of the selected fund indexes in the constructed evaluation model is firstly confirmed by professional opinion and DEMATEL algorithm. After that, ANP algorithm is used as the framework to get weights. Based on professional investors’ suggestion, Weighted FCM is used to classify fund according to its historical return and risk. This helps to provide investors a direction to choosing funds. Then, the study uses TOPSIS algorithm which is multiple attribute decision making skill to rank each fund, providing more detailed fund choosing guidance. The fund evaluation model is quite easy to understand for investors. As long as investors know what kind of investors they belong to, they can quickly get suggestions regarding fund choosing. Also, they can calculate their preference to return and risk indexes, which is a percentage for them to consider when making individual fund choosing decisions.