ARMA模型参数估计算法改进及在股票预测中的应用

Improved Algorithm of Parameter Estimating for ARMA Model and Its Application in the Stock Markets Forecasting

针对股票市场中ARMA模型的识别、建立和估计问题,提出一种ARMA模型参数估计的改进算法,以加快计算的收敛速度和提高模型参数估计的精确度.该算法借助反向过程确定初值,结合优化阻尼最小二乘法求解模型参数.应用该算法对预测股票价格进行了仿真试验,并与SAS预测结果作了对比,获得了满意的效果.实验结果表明该算法在预测性能上有了较大的提高,证实了该算法的有效性.

Aimed at the problems of identifying, establishing and estimating of ARMA model in the Stock Market Forecasting, the paper proposes a sort of improved algorithm of parameter estimate for ARMA model so as to expedite the computing convergence speed and enhance the accuracy of model parameter estimate. By means of backward process, it determines the initial value, and combined with optimized damping least- squares method, it solves model parameters. It carries out the simulation experiment to forecast stock price, and makes comparison with forecast result of SAS showing that the satisfactory result is obtained. The experiment shows that the forecasting performance of this algorithm is greatly improved, validating its effectiveness.