本研究結合最適化投資組合與型態投資，將個股的帳面市價比、資產報酬率兩個特徵型態經過產業調整後，依照資產報酬率、帳面市價比的順序進行相依分組，計算得到的型態投資報酬作為最適化投資組合的參數估計，研究樣本為市值前五十大的公司。研究發現型態最適投資組合比Markowitz 方法還要高了5%的年化超額報酬，也比市場平均每年多了將近8%的超額報酬，2005 年後的結果尤其顯著。進一步考慮交易成本和投資個股上限後，並沒有影響本文的結果。最後，我們發現投資組合的優異表現主要是來自於型態投資報酬改善了期望報酬的估計。;In this study, I combine the concept of portfolio optimization and style investing. I use style returns to estimate expected returns and covariance matrix. For calculating style returns, portfolios are formed by dependently quintile sorting on ROA and BM demeaned by industry. The research sample are top 50 largest market capitalization Taiwan stocks. I find that style portfolio optimization can beat traditional portfolio optimization and market by 5% and 8% annualized excess returns on average. The result does not change after considering trading costs and the limit of investing. The outstanding result is mainly contributed by improving the estimation of expected returns.
