指數股票型基金最適定期式計量投資模式之探討－－以台灣五十ETF為例

A Study on the Optimal Dollar-cost Averaging Quantitative Investment Model for Exchange Traded Funds-Evidences from Taiwan 50 ETF

由於欠缺經濟景氣與證券市場轉折變化之分析能力，因此，追求與市場績效同步的指數股票型基金（ETF）即成為現今眾多投資人的替代選擇，本研究即以ETF投資人的立場試圖建構出最佳投資策略。先前多數研究指出相較於單筆投資，定期式投資策略確可發揮分散投資時點、平均投資成本及降低投資風險等功效，因而未來極可能推廣至ETF。本研究妥適結合多種市場技術面指標與經濟基本面指標，並以此雙重層面綜合指標做為調整各期投資金額之依據，藉以建構出類似之定期不定額投資策略，並從實證分析中篩選出最適之定期式計量投資模式。本研究並選取台灣五十ETF做為實證對象且獲致兩項主要發現，首先，本研究所提出之定期不定額投資策略之績效確實相當幅度超越傳統定期定額策略。其次，經過多面向指標建構及實證比較分析後，確可篩選出績效表現相對優異之定期式計量投資模式。

The stock-type exchange traded funds (ETFs) whose performances are tracked and simulated to be consistent with the given market index have been chosen as an alternative by general investors because of failing to effectively manage the fluctuation of economic situation and security market for lack of analysis ability. This study focuses on creating the optimal investment strategies in terms of ETF investors. Comparing with the lump-sum strategy, the majority of the related researches indicate that the dollar-cost averaging strategy can realize the effectiveness of dispersing the investment timing, averaging investment cost and reducing investment risk. Consequently, the business is likely popularized to ETFs. A variety of market technical indicators and economic fundamental indicators are sensibly combined. Adjustment in periodical investing amount is determined based on the value of bi-aspect integrated indicator at that time. As a result, an investment strategy similar to variable amount dollar-cost averaging is created. Furthermore, the optimal quantitative investment model can be found out through empirical analysis. This study takes Taiwan 50 ETF as the empirical study sample. Two main findings are acquired from empirical results. First, the variable amount dollar-averaging strategy developed by this study, overall, is exactly considerably superior to traditional dollar-averaging strategy. Second, this study definitely creates an optimal dollar-cost averaging quantitative investment model that can yield the relatively outstanding performance via the multi-feature analysis and comparison.