股票预测对于投资者和金融市场具有重要意义,而同时刻画股票数据蕴含的非线性、非对称特性并解释外部冲击和复杂动态及状态变化的影响,对于传统GARCH模型来说存在局限性。因此为解决上述问题,本文结合非线性势GARCH (NPGARCH)和隐马尔可...股票预测对于投资者和金融市场具有重要意义,而同时刻画股票数据蕴含的非线性、非对称特性并解释外部冲击和复杂动态及状态变化的影响,对于传统GARCH模型来说存在局限性。因此为解决上述问题,本文结合非线性势GARCH (NPGARCH)和隐马尔可夫(HMM)两模型优势构建基于隐马尔可夫的带有外生变量的双非GARCH (HMM-DNGARCH-X)模型对股票收益率进行预测。首先根据HMM将股票价格波动划分为正常、异常状态,通过Baum-Welch算法估计模型参数,并采用Viterbi算法对隐状态序列进行识别;随后,将不同状态对应的收益率带入到HMM-DNGARCH-X模型进行预测分析。通过模拟研究,对模型的有效性进行了验证。基于上证国债指数的实证分析结果表明,与NPGARCH、GJR-GARCH、DNGARCH等模型相比HMM-DNGARCH-X模型的拟合损失均较小,并且预测效果显著优于其他模型。因此,所述HMM-DNGARCH-X模型能够较好地预测股票价格波动情况,在金融投资事件中具有广泛的应用潜力。Stock prediction is vital for investors and financial markets. Traditional GARCH models have limitations in capturing nonlinearity, asymmetry, and external shocks. To address this, we developed the HMM-DNGARCH-X model by combining the strengths of both models. Initially, HMM classifies stock price movements into normal and abnormal states, estimates parameters with the Baum-Welch algorithm, and identifies hidden states using the Viterbi algorithm. Returns from different states are then input into the HMM-DNGARCH-X model for prediction. Simulations validate the model’s effectiveness. Empirical analysis on the Shanghai Treasury Bond Index shows that HMM-DNGARCH-X has lower fitting losses and better predictive performance than NPGARCH, GJR-GARCH, and DNGARCH models. Thus, HMM-DNGARCH-X demonstrates significant potential for predicting stock price volatility in financial investments.展开更多
本研究运用DCC-GARCH模型,针对中国八大证券公司在2015年7月1日至2023年7月1日期间的股票收益率数据进行了深入分析,旨在揭示这些证券公司间系统性风险的动态关联性。实证分析结果显示,中国这八大证券公司的股票收益率呈现出波动聚集性...本研究运用DCC-GARCH模型,针对中国八大证券公司在2015年7月1日至2023年7月1日期间的股票收益率数据进行了深入分析,旨在揭示这些证券公司间系统性风险的动态关联性。实证分析结果显示,中国这八大证券公司的股票收益率呈现出波动聚集性,并且风险动态相关性为正。鉴于此,证券公司应持续强化其对系统性风险的管理能力,不断完善内部风险防控体系,增强其风险抵御力,以期有效降低风险对公司造成的损失。This paper establishes DCC-GARCH model to study the stock returns of eight major securities companies from July 1, 2015 to July 1, 2023, and finds the dynamic correlation of systemic risk among eight major securities companies in China. Through empirical analysis, it can be found that the stock returns of China’s eight securities companies have the characteristics of volatility clustering, and there is a positive risk dynamic correlation. Securities companies should continuously strengthen their ability to manage systemic risks, adhere to improving their internal risk prevention and control mechanism, and improve their ability to resist risks, so as to reduce the losses brought by risks to securities companies.展开更多
文摘股票预测对于投资者和金融市场具有重要意义,而同时刻画股票数据蕴含的非线性、非对称特性并解释外部冲击和复杂动态及状态变化的影响,对于传统GARCH模型来说存在局限性。因此为解决上述问题,本文结合非线性势GARCH (NPGARCH)和隐马尔可夫(HMM)两模型优势构建基于隐马尔可夫的带有外生变量的双非GARCH (HMM-DNGARCH-X)模型对股票收益率进行预测。首先根据HMM将股票价格波动划分为正常、异常状态,通过Baum-Welch算法估计模型参数,并采用Viterbi算法对隐状态序列进行识别;随后,将不同状态对应的收益率带入到HMM-DNGARCH-X模型进行预测分析。通过模拟研究,对模型的有效性进行了验证。基于上证国债指数的实证分析结果表明,与NPGARCH、GJR-GARCH、DNGARCH等模型相比HMM-DNGARCH-X模型的拟合损失均较小,并且预测效果显著优于其他模型。因此,所述HMM-DNGARCH-X模型能够较好地预测股票价格波动情况,在金融投资事件中具有广泛的应用潜力。Stock prediction is vital for investors and financial markets. Traditional GARCH models have limitations in capturing nonlinearity, asymmetry, and external shocks. To address this, we developed the HMM-DNGARCH-X model by combining the strengths of both models. Initially, HMM classifies stock price movements into normal and abnormal states, estimates parameters with the Baum-Welch algorithm, and identifies hidden states using the Viterbi algorithm. Returns from different states are then input into the HMM-DNGARCH-X model for prediction. Simulations validate the model’s effectiveness. Empirical analysis on the Shanghai Treasury Bond Index shows that HMM-DNGARCH-X has lower fitting losses and better predictive performance than NPGARCH, GJR-GARCH, and DNGARCH models. Thus, HMM-DNGARCH-X demonstrates significant potential for predicting stock price volatility in financial investments.
文摘本研究运用DCC-GARCH模型,针对中国八大证券公司在2015年7月1日至2023年7月1日期间的股票收益率数据进行了深入分析,旨在揭示这些证券公司间系统性风险的动态关联性。实证分析结果显示,中国这八大证券公司的股票收益率呈现出波动聚集性,并且风险动态相关性为正。鉴于此,证券公司应持续强化其对系统性风险的管理能力,不断完善内部风险防控体系,增强其风险抵御力,以期有效降低风险对公司造成的损失。This paper establishes DCC-GARCH model to study the stock returns of eight major securities companies from July 1, 2015 to July 1, 2023, and finds the dynamic correlation of systemic risk among eight major securities companies in China. Through empirical analysis, it can be found that the stock returns of China’s eight securities companies have the characteristics of volatility clustering, and there is a positive risk dynamic correlation. Securities companies should continuously strengthen their ability to manage systemic risks, adhere to improving their internal risk prevention and control mechanism, and improve their ability to resist risks, so as to reduce the losses brought by risks to securities companies.
