基于马尔科夫模型的期货市场收益率波动预测研究

Research on the Futures Market Yield Fluctuation Prediction Based on Markov Model

随着全球经济一体化,仅对收益率进行预测已不足以应对日益复杂的市场风险,若能对金融市场的波动性进行准确预测,可抑制投资者在规避风险和进行套期保值时的过度投机和操纵市场行为。本文以大豆2号期货市场和黄金期货市场作为研究对象,比较不同分布下GARCH类模型对大豆2号与黄金期货市场波动的刻画能力,选择最优模型预测大豆2号期货和黄金期货的波动率;采用隐马尔科夫模型(Hidden Markov Model,HMM)确定和预测期货市场的波动状态;采用不同的损失函数对EGARCH模型及组合模型进行评估,检验模型预测的准确性。实证研究表明,大豆2号和黄金期货的残差更服从于正太分布,GARCH(1,1)对期货序列的拟合更有效;HMM-EGARCH模型对大豆2号与黄金期货市场波动率的预测比其他模型更加准确。

With the integration of the global economy,forecasting yields alone is no longer enough to deal with increasingly complex market risks.If the volatility of financial markets can be accurately forecasted,it can restrain investors from risk aversion and hedging.excessive speculation and market manipulation.This paper takes soybean No.2 futures market and gold futures market as the research objects,compares the ability of GARCH models to describe the volatility of soybean No.2 and gold futures markets under different distributions,and selects the optimal model to predict the volatility of soybean No.2 futures and gold futures.;Use Hidden Markov Model(HMM)to determine and predict the fluctuation state of futures market;use different loss functions to evaluate the EGARCH model and the combined model to test the accuracy of the model prediction.Empirical research shows that the residuals of soybean No.2 and gold futures are more subject to normal distribution,and GARCH(1,1)is more effective for fitting the futures sequence;HMM-EGARCH model predicts the volatility of soybean No.2 and gold futures markets more accurate than other models.