摘要
采用GARCH模型和BP神经网络模型,利用上海A股30支股票(6类,每类各5支)2015年6月29日至2017年6月30日的日收盘价,分别进行短期(2017年7月3日至7日)、中期(2017年8月14日至18日)和长期(2017年9月25日至29日)预测.结果表明:在短期预测中,无论是否考虑当日价格波动对预测结果的影响,2个模型的每日与一周总体预测效果的差异均不具有统计学意义(P>0.05);在中期预测中,无论是否考虑当日价格波动对预测结果的影响,BP神经网络模型的每日与一周总体预测效果均优于GARCH模型,且总体差异具有高度统计学意义(P<0.01),不考虑当日价格波动时每日预测效果的差异具有统计学意义(0.01<P<0.05),考虑价格波动时每日预测中有3日预测效果的差异不具有统计学意义(P>0.05),有2日预测效果的差异具有统计学意义(0.01<P<0.05);在长期预测中,无论是否考虑当日价格波动对预测结果的影响,BP神经网络模型的每日与一周总体预测效果均优于GARCH模型,且总体差异具有高度统计学意义(P<0.01),每日预测效果的差异不具有统计学意义(P>0.05);随着预测周期的延后,预测误差逐渐增大.
The GARCH model and the BP neural network model are used to make short-term(July 3 to 7,2017),mediumterm(August 14 to 18,2017)and long-term(September 25 to 29,2017)forecasts using the daily closing prices of 30 stocks(6 categories,5 each)of Shanghai A shares from June 29,2015 to June 30,2017.The results show that in the short-term forecast,whether or not considering the impact of daily price fluctuations on the forecasting results,the differences of the two models on the daily and weekly overall forecasts are not statistically significant(P>0.05).In the medium-term forecast,whether or not considering the impact of daily price fluctuations on the forecasting results,the daily and weekly overall forecasts of the BP neural network model are better than those of the GARCH model,and the difference of overall forecast is highly statistically significant(P<0.01).The differences of daily forecasts effect are statistically significant when the price fluctuations of the day are not considered(0.01<P<0.05),and when price fluctuations are considered,the differences on 3 days forecasts of the daily forecasts are not statistically significant(P>0.05)while the differences on the other 2 days forecasts are statistically significant(0.01<P<0.05).In the long-term forecast,whether or not considering the impact of daily price fluctuations on the forecasting results,the daily and weekly overall forecasting effects of the BP neural network model are better than those of the GARCH model.The difference of overall forecast is highly statistically significant(P<0.01),and the differences of daily forecasts are not statistically significant(P>0.05).Besides,the forecasting error gradually increases with the delay of the forecasting period.
作者
崔文喆
李宝毅
于德胜
CUI Wenzhe;LI Baoyi;YU Desheng(College of Mathematical Science,Tianjin Normal University,Tianjin 300387,China)
出处
《天津师范大学学报（自然科学版）》
CAS
北大核心
2019年第5期30-34,共5页
Journal of Tianjin Normal University：Natural Science Edition
基金
国家自然科学基金资助项目(11271046
11671040)
天津师范大学博士基金资助项目(52XB1414)
