依據股票特性的投資已經經過長時間的討論及研究,但是究竟用什麼方式或模型去捕捉股票特性的移轉仍然是一個問題,許多的研究者已經提出過許許多多不同的策略去做嘗試。在本文中,我會去重新檢驗這些已經被提出來的模型以及一些新的模型運用,例如羅吉斯迴規模型、多項式羅吉斯迴歸模型、鑑別分析、變異數極小最適化因子模型、動能與反向策略。至於在解釋因子的部分,我會使用三種類型的因子,分別是財務因子、市場狀況因子以及落後期的標的資產報酬。而在分類所依據的股票特徵部分,我會使用規模特徵、股價淨值比、本益比、現金流量對股價比、股票殖利率。我認為因為模型所必須估計的因子以及對於整個市場的代表性,使得模型中所涵蓋的標的資產以及因子數量對研究者來說是一種取捨的觀念。所以我嘗試想要取得一個衡平的資產及因子數量來極大化模型的估計能力,並利用上述的模型及因子作出估計,而估計的結果也令人振奮。總結來說,羅吉斯迴規模型、鑑別分析、變異數極小最適化因子模型、動能策略都在某種程度上捕捉了股票特性轉換的軌跡。 Style rotation was discovered for a long time. But practically how to combine forecasting methodology with style rotation is still a dispute. Many researchers propose different kinds of style switching strategies. In my article, I examine the performance of several alternative style rotation strategies, including logit model, multinomial logit model, discriminant analysis, factor-based mean-variace optimization, momentum and contrarian strategies. For explanatory factors, I employ financial variables, market condition variables and VAR(1). For firm characteristics, I employ size, BM, earnings to priceratio, cash flow to price ratio and dividend to price ratio. The empirical results indicate that the logit regression model, discriminant analysis, mean-variance optimization and momentum strategy all accurately take advantage of style rotation patterns to some extent.