基于贝叶斯的金融时间序列预测研究

On the Financial Time Series Prediction Based on Bayesian

传统的时间序列分析与预测方法没有考虑样本和参数的先验信息,导致预测结果和实际数据的偏差较大,贝叶斯参数估计方法可以充分利用参数的先验信息,使得估计参数的方差更小,估计结果更加精确,预测结果更真实有用。随着MCMC方法和WinBUGS软件的发展,贝叶斯分析方法估计模型的计算困难逐渐减弱,因此,近年来贝叶斯时间序列预测方法越来越受到关注。本文基于上证指数收盘价的数据,采用Eviews和WinBugs软件,对样本数据进行预处理,利用贝叶斯参数估计方法进行时间序列自回归模型的实证研究分析。

The traditional time series analysis and prediction method does not take the prior information of the samples and parameters,which leads to the large deviation between the prediction results and the actual data.However the Bayesian method can make full use of the prior information of the parameters,so that the variance of estimated parameters is much small,the results are more accurate,the forecast is more realistic and useful.With the development of MCMC method and WinBUGS software,the Bayesian time series prediction method has been paid increasingly attention in recent years.Based on the data of Shanghai Composite Index,this paper used Eviews and WinBugs softwares to preprocess the sample data and analyze the time series autoregressive model by Bayesian parameter estimation method.