本研究選取十四家金融控股公司為研究樣本，使用GARCH波動性修正歷史模擬法之風險值，以報酬風險來做為金融控股公司間績效評估，供投資人在諸多層面的決策參考。主要結論如下：（一）十四家金融控股股價報酬具有異質性，可使用GARCH波動模型。（二）以HW法算出之風險值以新光金控最低，富邦金控最高，而透過三種誤差衡量指標，一致發現富邦金控具有較佳的預測能力。（三）以風險報酬來計算金融控股公司之績效，以玉山金控所代表小而美的金控公司較佳，而大型的金控公司表現則不如預期。造成此原因之一；金控公司成立迄今才一年多，小型金控公司磨合期較快，對風險能較有效的控管；而大型金控公司則是產業繁雜，需要較長的磨合期，綜效無法一下表現出來，對風險也需較長的時間才能掌控。 對投資人的建議於投資時若背景與本研究相似，可以參照本研究的實證結論來做參考，依波動性修正的歷史模擬法所預測的風險值當做決策的指標。配合歷史報酬來參考進場時機；對後續研究者建議，因本研究僅針金融類股中的金控進行股價報酬風險的績效評估分析，後者可以再針對電子類股進行探討，由於電子類股波動明顯且為投資大眾所喜好，針對電子類股進行實證分析，將獲得更顯著的結果，亦可與金融類股進行比較，以得知是否對類股有不同的實證結果，在方法上更可使用不同的模擬的風險值來做比較，以求更精確的實證分析。

This article choose the fourteen Finance holding company stock to study, by using GARCH model to adjust the Historical Simulation will increase the feasible of forecasting, and thus it can be applied to stock selection to forecast the future performance. It will make effort investors to decide. The conclusion in my study are: (1) The empirical results show that Finance holding company stock price had volatility clustering and heteroscedasticity to fit GARCH model. (2) using HW methods to estimate value at risk, finding shin-kong holding company is the lowest, fubon holding company is the highest, and fubon holding company is the best predict. (3) using risk/return to calculate the performance of Finance holding company, e-sun holding company which present small Finance holding company is the best, and the other big Finance holding company performance is not good enough. The reason is small Finance holding company united quickly, and control risk efficiency. It is suggested the investor that consider of this result if their background is similar with this study, and depend on HW methods to decide. For researchers, they can use electric stock. Maybe the performance of electric stock is better than Finance holding company.