BP神经网络与ARMA-GARCH模型在股票预测中的对比分析  

Comparative analysis of stock forecast based on BP neural network and ARMA-GARCH model

作　　者：张如梦 张华美 ZHANG Rumeng;ZHANG Huamei(School of Mathematics and Statistics,Xinxiang University,Xinxiang 453003,China;School of International Economics,Shanxi Institute of International Trade&Commerce,Xi′an 712046,China)

机构地区：[1]新乡学院数学与统计学院,河南新乡453003 [2]陕西国际商贸学院国际经济学院,陕西西安712046

出　　处：《高师理科学刊》2021年第5期14-20,共7页Journal of Science of Teachers＇College and University

基　　金：新乡市政府决策研究招标课题(B20165);陕西国际商贸学院校级项目(SMXY201905)。

摘　　要：考虑到股票价格瞬息万变,波动性强,用BP神经网络与ARMA-GARCH模型对上汽集团收盘价预测.对BP神经网络与ARMA-GARCH模型的股票预测结果进行对比分析,结果表明,在预测未来20 d收盘价时,BP神经网络平均绝对误差比ARMA-GARCH模型低31.4%;在预测未来6 d收盘价时,ARMA-GARCH模型平均绝对误差比BP神经网络低7.4%.说明BP神经网络在长期预测中更为精准,而ARMA-GARCH模型在短期预测中具有微弱优势.In view of the fast changing and volatility of stock price,BP neural network and ARMA-GARCH model was used to predict the closing price of SAIC Group.The stock prediction results of BP neural network and ARMAGARCH model are compared and analyzed,the results show that the average absolute error of BP neural network is 31.4%lower than that of ARMA-GARCH model in predicting the closing price in the next 20 days,the average absolute error of ARMA-GARCH model is 7.4%lower than that of BP neural network in predicting the closing price in the next 6 days.It shows that BP neural network is more accurate in long-term prediction,while ARMA-GARCH model has a weak advantage in short-term prediction.

关 键 词：股票价格 BP神经网络模型 ARMA-GARCH模型 

分 类 号：O29[理学—应用数学] F832[理学—数学]