摘要
期货市场波动性反映了市场的活跃度和流动性,是政府管控市场的重要决策来源,是投资者测量风险、实现资产保值的有利工具。已有研究表明,因加入了对过去时期的预测方差,GARCH模型比ARCH模型更能反映市场数据信息。然而在实际应用中,GARCH模型经常因数据的离散性而无法适应金融市场的结构突变,进而导致波动性预测效果不够理想。为解决上述问题,结合HMM和GARCH模型预测中国期货市场收益率的波动性。通过GARCH模型计算期货的波动率序列;利用K均值法对波动率序列聚类得出观察序列;根据HMM划分波动率的状态,将不同状态对应的收益率代入HMM-GARCH模型,以得到不同状态下的波动率;通过VIX公式计算波动指数,以测量市场的波动性。基于以上逻辑,选用沪深300股指期货作为标的,以2015年5月至2016年4月为样本期,验证模型的有效性。研究结果表明,一方面,HMM-GARCH模型的MAD和MSE两种损失函数值均比GARCH模型的低,表明拟合损失和错误少,可见与GARCH模型相比,HMM-GARCH模型能更好地拟合样本数据并预测市场信号;另一方面,基于HMM-GARCH模型的波动率指数显示,在样本期内沪深300股指期货由前期的小幅频繁波动转为大幅跳跃性波动,此后波动幅度保持较高水平并呈现增长态势,最终继续转为大幅跳跃性波动,与样本期内沪深300股指期货价格的实际波动态势一致。因此,所述HMM-GARCH模型能够较好地测量中国期货市场波动状况,反映期货投资者对未来中国期货市场的预期。同时,能够为政府设置金融衍生品定价提供决策依据,为投资者测量市场风险、择取投机策略、合理配置资产提供客观的量化指标,有助于培养投资者的投资理性,促进中国期货市场的繁荣稳定发展。
The volatility of a futures market reflects the activity and liquidity of the market.For governments,it is a vital source of decision-making to control the market;while for investors,it is a useful tool to measure risks and achieve asset maintenance.Previous studies have shown that the GARCH(generalized auto regressive conditional heteroskedasticity)model is more responsive to market data than ARCH(auto regressive conditional heteroskedasticity)model because of the inclusion of the prediction variance over the past period.In practice,however,the GARCH model often fails to adapt to the structural changes in the financial market because of the discreteness of the data,which leads to the inefficiency of the volatility prediction effect.To solve the above problems,this paper attempts to predict the volatility of Chinese futures market returns by combining HMM(hidden Markov model)and GARCH.Firstly,this paper calculates the volatility series of futures by GARCH model.Then it clusters the volatility series to obtain the observation sequence through K-means method.After that,it divides the state of volatility according to the HMM model,and obtains the volatility under different conditions by substituting the corresponding rates of different states into the HMM-GARCH model.Finally,it calculates the volatility index using the VIX(volatility index)formula.Followed by the above logic,this paper selects the CSI 300 stock index futures as the experimental subject,and chooses May 2015 to April 2016 as the sample period to verify the validity of the model.In the experiment results,the two loss function values(MAD:mean squared difference and MSE:mean squared error)of the HMM-GARCH model are both lower than those of the GARCH model,indicating that the fitting loss and errors of the HMMGARCH model are less.Compared with the GARCH model,the HMM-GARCH model can better fit sample data and predict market signals.In addition,according to the forecasting of volatility index,the price of CSI 300 stock index futures changed from a small amount of frequent volatility in the previous period to a large jump in volatility.Since then,the volatility remained at a high level and had a growth trend.Eventually,it continued to turn into sharp jumps.Such fluctuation of CSI 300 stock index futures is consistent with its actual fluctuation.Therefore,the HMM-GARCH model described in this paper can better measure the volatility of Chinese futures market and reflect the expectations of futures investors for future domestic futures markets.At the same time,it can provide the basis for the government to set the pricing of financial derivatives,provide objective quantitative indicators for investors to measure market risk,choose a speculative strategy,and rationally allocate assets,which will help cultivate investors’rational investment and promote the prosperity and stability of Chinese futures market.
作者
景楠
吕闪闪
江涛
JING Nan;LYU Shanshan;JIANG Tao(SHU-UTS SILC Business School,Shanghai University,Shanghai 201899,China;School of Economics and Management,Harbin Institute of Technology,Shenzhen,Shenzhen 518055,China)
出处
《管理科学》
CSSCI
北大核心
2019年第5期152-162,共11页
Journal of Management Science
