基于微分信息的ARMAD-GARCH股价预测模型

ARMA-GARCH模型进行股票价格收益预测时,只考虑了滞后历史数据所包含的信息,而对于在每个滞后时间点的变化趋势信息却未纳入计算模型进行统一考虑,在一定程度上影响了模型分析时序数据时的泛化能力.本文提出了一种基于微分信息的ARMAD-GARCH模型,在包含传统ARMA-GARCH模型对因变量的滞后值以及残差滞后值进行线性回归的基础之上,又在条件均值方程中增加了因变量滞后值的近似微分信息,用以融合股票价格变化趋势信息,提高预测模型对于价格演变方向的判别能力.通过对于不同市场综合股指收益率数据的实证研究表明,ARMADGAR,CH模型在数据除噪,趋势判别以及预测精确度等方面均优于一般的ARMA-GARCH模型.

A differential-information based ARMAD-GARCH stock price forecasting model

ARMA-GARCH model only takes historical data into consideration when predicting the stock price return, without taking into account the trend information at each belated time point, which, to some degree, influences the model's generalization capability. In this paper, a differential-based ARMAD-GARCH is developed by incorporating approximate differential signal of the belated dependent variable into the traditional ARMA-GARCH model's mean equation, which is used to integrate the stock price trend information and improve the model's discrimination capacity for price's evolution direction. A empirical study, based on different composite index returns, demonstrates that ARMAD-GARCH model is superior to the traditional model in terms of data denosing, trend discrimination and prediction accuracy.