论文部分内容阅读
本文对郑州期货糖0809主力合约的价格首先进行多元线性回归,探求其与纽约期货糖价和郑州现货糖价的关系,进而在发现回归残差具有周期效应的基础上进行时间序列的频域分析,并同时考虑各种突发事件的影响,在模型中加入示性变量进行适当修正,经过ADF单位根检验确定此时的残差已为平稳序列之后建立ARMA模型,并接受最终残差为白噪声.将上述分解过程进行整合,估计模型系数并剔除其中的不显著变量便得到最终的拟合方程,在此基础上对后续三天的郑州期货糖价进行动态预测,结果显示真实价格均落在所给95%置信区间内.“,”In this paper, we firstly established a multivariate linear regression to analyze the relationship among the future sugar price in 0809 contract in Zhengzhou Futures Exchange with the future sugar price in New York and the spot price of sugar in Zhengzhou. Then, the spectral analysis was applied to explore the cycle effect and indicator variables were added to explain the unexpected factors which demonstrated that our model was reasonable and applicable. After that, we adopted ARMA model based on the residuals which were stationary through ADF unit root test. Through all these steps, we built up the final model by estimating the coefficients and eliminating the insignificant variables. By our model, the future sugar prices of the following three days were predicted and it was showed that the real future sugar prices were exactly in 95% confidence intervals.