基于ARIMA模型的股市价格规律分析与预测

Analysis and Forecast of Stock Price Law Based on ARIMA Model

摘要: 股市价格涉及诸多不可控因素，各个因素之间关系错综复杂。本文以汉王科技股票2019年6月1日~9月21日的4种交易价格OP，FP，SP，LP为例进行分析与预测。首先对其变化规律和特征进行了描述，然后利用最小二乘法，定量分析了4种交易价格与时间t的依赖关系。其次对非平稳时间序列进行d阶差分运算，化为平稳时间序列，再分别求出平稳时间序列的自相关系数ACF和偏相关系数PACF，通过对自相关图和偏相关图的分析得到最佳阶数q和阶层p，构建了ARIMA模型。最后对所建的ARIMA (p, d, q)模型进行残差检验，并利用检验通过的模型进行预测，结果显示所建模型较为简单、精确且拟合效果较好。

Abstract: The price of stock market involves many uncontrollable factors, and the relationship among them is complex. This paper takes the four transaction prices of Hanwang Technology Stock (OP, FP, SP, LP) from June 1 to September 21, 2019 as example to analyze and forecast. Firstly, this paper described the law and characteristics of price change, and then quantitatively analyzed the dependence of four kinds of transaction prices on time by using the least square method. Secondly, using d-order difference operation to make the non-stationary time series transformed into stationary time series. After the calculation of the autocorrelation coefficient ACF and partial correlation coefficient PACF of stationary time series, through the analysis of autocorrelation graph and partial correlation graph, we get the best order q and level p, and construct ARIMA model. Finally, use the ARIMA (p, d, q) model which is passed the test for prediction. The results show that the model is simple, accurate and the fitting effect is good.