近年來大陸股市的變化對於世界金融環境的影響力加劇,大陸方面的成長力近年受到矚目,隨兩岸金融合作逐漸開放,諸多金融商品陸續出現。因此本研究試圖探討各種波動性模型與條件分配對於滬深300指數期貨高頻資料之波動度影響。研究期間包含2013年4月1日至2014年1月30日之5分鐘高頻資料,其中以2013年4月1日至2013年9月30日作為樣本內估計,取2013年10月8日至2014年1月30日進行樣本外預測。試圖在常態分配、T分配及GED分配下進行ARCH、GARCH及EGARCH模型配適度分析,並以RMSE、MAE及泰爾不等係數衡量預測績效。分別在預測模型的各種損失函數下找出配適性好的模型,研究發現殘差分配設定較模型設定更重要,透過選定適當的殘差分配能夠提高模型波動性的預測能力。因此針對不同金融商品研判適當的條件分配,有助於提高預測能力,能夠幫助我們進行風險管理。
The change of the stock market of Mainland China intridges the more impact on the world’s financial environment. Likewise, the rate of growth has become a focus worldwide. There are variety of financial products because of the gradually opening of cross-strait financial cooperation. For this reason, this reaserch trys to focus on how the variety of volatility models and codition distributions affect the Shanghai and Shenzhen 300 index futures with high frequency. The time period of the research which begins on the date of April 1, 2013 and ends on January 31, 2014. In this time interval, the author trys to divide it in two parts: One is named for the in- the-sample estimator whose time interval is from April 1, 2013 to September, 30, 2013; the other is out-the-sample forecast whose time period is from October 1, 2013 to January 30, 2014. Moreover, the researcher aims at implementing the fitted analysis of ARCH, GARCH, and EGARCH models under Normal, T, and GED distributions, also using RMSE, MAE, and Theil inequality to be the measurement of performance forecasting. The result of this study shows that setup for residual distribution is more important than that for models, which means choosing the fitted residual distribution to elevate the estimate ability of model’s volatility. Thus, find out the most fitted distribution for different financial products helps to get higher ability for forecasting, and helps us to do the risk management decision.