股票市场快速发展,股票价格波动性研究备受关注,准确预测股价走势对投资者决策和市场稳定意义重大。鉴于股票价格波动的不确定性与非线性特征,单一模型预测效果欠佳。为此,本文提出将GARCH与BP神经网络相结合的组合预测方法,以中国农业银行股票日收盘价数据为例,基于误差修正思想构建组合模型,运用BP神经网络对GARCH模型的残差数据进行预测校正。研究结果表明组合模型预测效果优于单一模型,验证了该组合模型在提高股票价格预测准确度方面的有效性。With the rapid development of the stock market, the study of stock price volatility has attracted much attention, and accurate prediction of stock price movements is of great significance to investors’ decision-making and market stability. In view of the uncertainty and nonlinear characteristics of stock price volatility, the prediction effect of a single model is not good. For this reason, this paper proposes a combined prediction method combining GARCH and BP neural network, taking the daily closing price data of Agricultural Bank of China as an example, constructing a combined model based on the idea of error correction, and utilizing BP neural network to correct the residual data of the GARCH model for prediction. The results show that the combination model predicts better than a single model, which verifies the effectiveness of the combination model in improving the accuracy of stock price prediction.
随着金融体系的快速发展,商业银行作为金融体系中的重要组成部分,在推动整个行业稳定运行和发展的同时,面临更多的风险和挑战。市场风险已经成为商业银行面临的主要风险,本文以贵阳银行为例,采用贵阳银行每日收盘价数据建立股票价格的日对数收益率序列,根据金融时间序列的波动性和异方差性特征,以GARCH模型为基础建立反映其股价变化的波动率模型,计算VaR值。研究结果表明贵阳银行的VaR值高达0.216629,说明贵阳银行收益率在95%的置信水平上损失极限为资产市场价值的21.66%,面临着较大的市场风险。因此,贵阳银行应该采取相应的措施管理和应对面临的市场风险,确保银行的稳健运营和持续发展。With the rapid development of the financial system, commercial banks, as an important part of the financial system, are promoting the stable operation and development of the entire industry while facing more risks and challenges. Market risk has become the main risk faced by commercial banks. Taking Bank of Guiyang as an example, this paper uses the daily closing price data of Bank of Guiyang to establish a daily logarithmic return rate sequence of stock prices. According to the volatility and heteroscedasticity characteristics of financial time series, a volatility model reflecting its stock price changes is established based on the GARCH model to calculate the VaR value. The research results show that the VaR value of Bank of Guiyang is as high as 0.216629, indicating that the loss limit of the return rate of Bank of Guiyang is 21.66% of the market value of assets at the 95% confidence level, and it is facing relatively large market risks. Therefore, Bank of Guiyang should take corresponding measures to manage and deal with the market risks it faces to ensure the stable operation and sustainable development of the bank.
本文将函数系数GARCH-M模型推广到多元情形,研究了一类多元函数系数GARCH-M模型,旨在把序列之间的交互作用引入到风险厌恶的研究上。文章给出了函数系数和模型参数的估计方法。数值模拟的结果表明,该方法的估计效果良好。实证分析基于上证综合指数和深证综合指数的日收益率数据,研究结果表明,多元函数系数GARCH-M模型能够更好地拟合所考虑的数据。In this paper, the GARCH-M model of function coefficients is extended to multivariate cases, and a class of GARCH-M model with multivariate function coefficients is studied, aiming at introducing the interaction between sequences into the study of risk aversion. The estimation methods of function coefficients and model parameters are given in this paper. Numerical simulation results show that the proposed method is effective. The empirical analysis is based on the daily return data of Shanghai Composite Index and Shenzhen Composite Index, and the research results show that the multivariate function coefficient GARCH-M model can better fit the considered data.
本文以上证指数的日收益率为样本,对上证指数建立GARCH-VaR模型,比较不同分布假定下GARCH类模型对上证指数波动率的拟合效果,计算并检验上证指数VaR值的预测结果对实际损失的覆盖情况。分析的结果表明,TARCH模型与EGARCH模型更适合测度上证指数条件方差,且在t分布下,模型能够更好地反映上证指数收益率扰动项的分布特征。进一步,为克服ARMA-GARCH模型在中长期预测中出现的较大误差,使用ARIMA-LSTM模型结合GARCH类模型预测指数波动率,有效提高了GARCH-VaR模型的预测准度。最后,通过TARCH模型,初步检验了我国股市注册制全面推行对上证指数波动率所产生的影响,发现该政策的实施显著降低了上证指数的波动幅度。This article takes the daily return of the Shanghai Composite Index as a sample, establishes a GARCH-VaR model for the Shanghai Composite Index, compares the fitting effect of GARCH models on the volatility of the Shanghai Composite Index under different distribution assumptions, calculates and tests the coverage of actual losses by the predicted VaR value of the Shanghai Composite Index. The analysis results indicate that the TARCH model and EGARCH model are more suitable for measuring the conditional variance of the Shanghai Composite Index, and under the t-distribution, the model can better reflect the distribution characteristics of the disturbance term of the Shanghai Composite Index return. Furthermore, to overcome the significant errors in medium- and long-term forecasting caused by the ARMA-GARCH model, the ARIMA-LSTM model combined with GARCH class models was used to predict index volatility, effectively improving the prediction accuracy of the GARCH-VaR model. Finally, through the TARCH model, the impact of the comprehensive implementation of the registration system in China’s stock market on the volatility of the Shanghai Composite Index was preliminarily examined, and it was found that the implementation of this po
