摘要
文章以上证综指2007年2月27日至2008年5月14日共293个交易日的收盘价数据为研究对象,通过构造样本外波动性预测能力指标,分析比较了GARCH类模型和SV类模型在国内市场上的适用性。文章首先估计GARCH类模型和SV类模型的参数,其中SV模型参数的估计采用最新的马尔可夫链蒙特卡罗方法(MCMC方法),并由WINBUGS软件加以实现。然后通过构造样本外预测能力指标对GARCH类模型和SV类模型的样本外预测效果进行比较,得出在国内市场SV类模型的拟合和预测效果要好于GARCH类模型。
This paper chooses the daily trading data of Shanghai Composite Index from 2007/2/27 to 2008/ 5/14 as the object of research, and picks the data of the sub-period from 2007/2/27 to 2008/2/27 as the in-sample data, and others as out-sample data to be examined. By comparing the out-sample forecasting ability of different models, we concluded that the SV models are better than GARCH models as a whole in describing the Shanghai Stock market. First we used the new method of MCMC to examine the parameters of SV model with the data in-sample in the condition of WINBUGS, and then forecasted the conditional variance of SCI out of sample. Finally, we used the ability index of out-sample forecasting to evaluate the fitting effect and forecasting ability of different models, and found that the SV models ware better than GARCH models in describing the Shanghai Stock market as a whole.
出处
《浙江万里学院学报》
2009年第2期1-7,共7页
Journal of Zhejiang Wanli University
基金
浙江省自然科学基金(Y607504)