股价时间序列的分析与预测研究

发布时间：2018-04-29 05:26

本文选题：ARMA模型 + BP神经网络　；参考：《大连理工大学》2013年硕士论文

【摘要】：股票市场是金融市场的一个重要组成部分,股票的趋势与波动能够反映一个国家的政治、经济和社会状况,指导国家宏观调控。同时,随着经济的发展和人民生活水平的提高,越来越多的人开始把多余的资金用来做投资,股票就是最常见的投资方式之一。因此,股票预测是一个很重要的金融课题,它对于国家经济宏观调控和投资者买卖股票并规避风险以获得最大收益,都具有重要意义。然而,股票价格受到很多因素的影响,如政治、经济、公司状况、投资者心理等等,因此股票价格序列是一个复杂的非线性动态系统,很难准确预测。 ARMA模型是最常用的时间序列预测模型之一,它能很好的处理线性问题,BP神经网络模型是一种非线性模型,对非线性序列能得到更好的预测效果。本文首先介绍这两种用于股票预测的单一模型的基本原理,用MATLAB编程建模实现了这两种模型并对中国石油和微软公司两支股票的股价进行了拟合与预测。然而,由于股价序列的复杂性,使用单一模型不能达到很好的预测效果,故本文提出了两种股票预测组合模型,ARMA-BP组合模型和ARMA-BP-Markov组合模型。ARMA模型和BP神经网络分别用于处理股价序列的线性和非线性部分,Markov模型用于对预测结果进行修正,使其更加准确。实验结果表明,组合模型的预测结果优于ARMA模型或BP神经网络单一模型,且ARMA-BP-Markov(?)(?)合模型能够比ARMA-BP组合模型得到更准确的预测结果。
[Abstract]:The stock market is an important part of the financial market. The trend and fluctuation of the stock market can reflect the political, economic and social situation of a country and guide the national macro-control. At the same time, with the development of economy and the improvement of people's living standard, more and more people begin to use surplus funds to invest. Stock is one of the most common investment methods. Therefore, stock forecasting is a very important financial issue, which is of great significance to national macroeconomic regulation and control and investors to buy and sell stocks and avoid risk to get the maximum return. However, stock price is influenced by many factors, such as politics, economy, company condition, investor psychology and so on. Therefore, stock price sequence is a complex nonlinear dynamic system, which is difficult to predict accurately. ARMA model is one of the most commonly used time series prediction models. It can deal with linear problems and BP neural network model is a kind of nonlinear model. This paper first introduces the basic principles of these two single models for stock forecasting, and implements the two models by using MATLAB programming model, and simulates and forecasts the stock prices of two stocks of PetroChina and Microsoft. However, because of the complexity of stock price sequence, using a single model can not achieve a good prediction effect. In this paper, two stock forecasting models, ARMA-BP combined model and ARMA-BP-Markov combination model. ARMA model and BP neural network are proposed to deal with the linear and nonlinear partial Markov models of stock price sequence, respectively, to modify the prediction results to make them more accurate. The experimental results show that the combined model is superior to the ARMA model or the BP neural network model, and ARMA-BP-Markovo model. The combined model can obtain more accurate prediction results than the ARMA-BP combined model.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F830.91;F224;TP183

【参考文献】