財務學者對解釋橫斷面股票報酬有相當多的研究，有許多學者認為股票報酬可以被因子所決定，他們認為因素負荷量（factor loading）高，報酬就會較高，報酬是被共變異風險所決定的。此一傳統的定價模型稱為風險因子模型(factor models)。而Daniel and Titman認為股票報酬是決定於特徵模型而非因子模型。本文以美國股票市場為研究對象，研究期間1963年到2002年利用買入持有法（buy-and-hold）來檢定何種資產定價模型具有較好的股票報酬預測能力。 本文的實證結果顯示，一年區間預測時，因子模型和特徵模型預測股票報酬能力的比較，特徵模型的股票報酬預測能力比因子模型好。 但以五年區間預測來比較時，以五年區間預測時並不能看出那一個模型是比較好的預測模型，只能看出資本資產定價模型（CAPM）預測績效是很不佳的。 Financial economists have extensively studied the cross-sectional stock return. Some economists consider certain factors that determine stock return. They think if there is high factor loading then the return increases and is controlled by covariance. This traditional asset pricing model is known as the “risk factor model”. Daniel and Titman argue that the stock return is more closely determined by “Characteristic model” than the risk factor model. For this research article, American stock market data from 1963 to 2002 was used. A “buy-and-hold” method was used to examine which asset pricing model would best predict the stock return. We compared the predictions of the stock return made by the factor model and the characteristic model. The results demonstrate that over one year the prediction of the stock return made by the characteristic model is more accurate than the prediction that the factor model. Over five years, there was no determination as to which model would make a more accurate prediction, but it was observed that the one factor model （CAPM）made an inaccurate prediction of the stock return.
