長期以來CAPM在理論與實務運用上普遍獲得支持，但近年來愈來愈多的實證研究證明，CAPM是不完全的，使得現代財務理論受到了重大挑戰，而越來越多的財務學者認爲CAPM不符合現實，無法完全解釋資本資産的定價，且CAPM的異常現象又不斷地出現。因此，本研究以台灣上市公司股票市場為研究對象，利用現代財務學較具影響力的投資組合模型－Naive模型、CAPM、Fama and French三因子模型、四因子模型及特徵模型，先進行預期報酬率平均數和共變數的估計，接著在不可賣空的前提下，以1994年至2002年之每年6月底選取市值最大的50支股票作為投資標的，建構各模型的最適投資組合，並以夏普指數評估各模型的投資績效，實證研究影響台灣證券市場報酬率的因子或特徵。 實證結果得知，以夏普指標來評估各模型的投資績效，四因子模型的投資績效最佳，其次為三因子模型及特徵模型，且三種模型均顯著優於市場投組及等權投組，同時亦較Naive模型及CAPM為佳。 The Capital Asset Pricing Model（CAPM）has been used widely in academics and practice for a long time. However, a growing literature also documents some major asset-pricing anomalies that the realized cross-section stock returns are not consistent with the predictions of the CAPM. Many researchers find that the size, price-to-book ratio, and the momentum factors explain much of the cross-sectional stock returns unexplained by the CAPM. Therefore, this thesis compares the performance of optimized portfolios constructed based on various models, including the naive model, the CAPM, the Fama-French three-factor model, the four-factor model and the characteristics model. The empirical study is implemented using monthly returns of all firms listed on the TSE during July 1989 to June 2003. Our empirical result shows that the four-factor model, the Fama-French three-factor model, and the characteristics model perform better in explaining the cross-section of stock returns than the Naive model and the CAPM.
