利用多元尺度法建構風險結構視覺化之投資組合

Abstract

本研究利用多元尺度法將資產報酬率之間的差異視覺化，形成二維動態資產集合之構形圖，並根據此視覺化之圖形，觀察所設定的資產集合整體的資產間關聯性，藉以更深入瞭解金融市場的風險結構。一般是採用相關係數矩陣觀察資產間的相關性大小，利用多元尺度法所形成的資產集合構形圖可幫助我們直接以視覺化的角度一次性的觀察所有資產間的關係。我們以臺灣等九個主要的股票市場作為研究樣本，實證結果發現，該構形圖可有效幫助我們以簡單的策略選取資產而形成較低風險的投資組合。

As a tenet in investment, diversification concept may be the most important guideline for the practitioners and researchers. In order to construct a well-diversified portfolio, we want to realize the whole risk structure of the financial market. In this study, we use multidimensional scaling technique to visualize time-varying returns between risky asset behaviors. From this market map, we can investigate adequacies of asset allocation and explore the risk structure of the markets more. We find that the MDS map can help us view and interpret the whole assets at once, thus we can select assets to construct a lower risk portfolio. Empirical results on several major markets also revealed the effectiveness of this approach.