基于EMD的我国股市波动的宏观经济成因及股价预测研究

发布时间：2018-01-03 12:36

本文关键词：基于EMD的我国股市波动的宏观经济成因及股价预测研究　出处：《东北财经大学》2013年硕士论文　论文类型：学位论文

【摘要】：经过了二十多年的发展,中国的股票市场取得了巨大成就,但由于市场化过程中存在各种矛盾以及经济体制转轨阶段具有特殊性,中国的股市还不成熟,表现为股价具有异常的波动现象。判断证券市场是不是成熟的一个指标就是波动特征,并且波动性是市场价格行为的重要决定因素,所以考察股市波动特征,有助于理解股市的价格行为。深入研究股市波动的因素,有助于投资者进行理性投资,促进国民经济长期健康稳定的发展。股市作为国民经济重要的组成部分,为国民经济的发展而服务。股市波动和宏观经济波动具有密切的关系,但由于股市具有其自身的波动规律,所以会出现“股经背离”。研究出一套有效的影响股价指数波动的理论,有利于我国股票市场的法律规范和监管机制的建立,也有利于引导我国的宏观经济政策,从而促进资本市场更加科学化的发展。随着中国经济的快速发展及中国股市不断完善,人们越来越多的参与到股市中来,也就希望能够更加准确的预测股市。从实践上,通过其他经济变量有效地预测股指的变化及股市的波动,投资者便能够提前采取行动,进而规避风险,最大化投资利润。本文提出了基于EMD(经验模态分解法)分解的股价波动的宏观经济成因的分析方法。首先,运用EMD方法将股价序列和宏观经济变量序列分解成几个不同周期的分量,然后对其重新组合,应用协整理论、Granger因果检验等对组合后的低频分量进行研究,最后进行总结。因为股票价格序列是非线性和非平稳的时间序列,所以对股票价格进行精确预测有一定的难度。为了克服传统计量模型预测的局限性,人们应用人工神经网络、支持向量机及遗传算法等人工智能方法进行预测,经过试验表明,这些方法的确优于传统的计量模型。本文提出了一种基于EMD和BP神经网络的非线性组合预测的方法,该方法运用EMD技术将股票价格序列分解成若干个不同频率的分量,对这些分量,构建不同的神经网络模型然后分别进行预测,得到各序列预测值,再将各个预测值进行合成得到最终预测值。本文的实证研究表明,首先,中国的宏观经济的发展和股票市场基本上是一致的,股价指数一定程度上是能够反映我国宏观经济发展的整体水平及趋势的。工业增加值增速、货币增长率对股价有正的影响,而通货膨胀率,汇率及利率对股价有负的影响,该结论与理论基本上是一致的。另外,本文提出的股价预测方法较单一的人工神经网络模型能够更好的预测股票价格。
[Abstract]:After more than 20 years of development, Chinese stock market has made great achievements, but due to various contradictions and economic transition stage has the particularity in the process of marketization, Chinese stock market is not mature, shown as the stock price fluctuation is abnormal. Judge the securities market is not a mature indicator is the fluctuation characteristics, and volatility is an important factor to determine the market price behavior, so the investigation of stock market volatility, the price behavior is helpful to understand the stock market. The research of factors of stock market volatility, help investors rational investment, promote long-term healthy and steady development of the national economy.
The stock market as an important part of the national economy, and the service for the development of the national economy. The stock market volatility and macroeconomic fluctuations have a close relationship, but because the fluctuation of the stock market has its own, so there will be nothing deviation ". To research an effective effect of stock price index fluctuation theory, establishing legal norms for China's stock market and regulatory mechanism, but also conducive to guide China's macroeconomic policies, so as to promote the capital market more scientific development.
With the rapid development of economy and Chinese China stock market continues to improve, more and more people to participate in the stock market, also hope to be able to predict the stock market more accurately. In practice, effectively predict changes and stock market index by other economic variables fluctuations, investors will be able to take action in advance, to avoid risks, to maximize the profit of investment.
This article based on the EMD (empirical mode decomposition) analysis method of macroeconomic volatility decomposition of the causes. First, using the method of EMD sequence of stock price series and macroeconomic variables is decomposed into several components with different periods, and the re combination, using cointegration theory, Granger causality test and other research on low frequency components after the combination, finally carries on the summary.
Because the stock price series is nonlinear and non-stationary time series, so it is difficult to accurately forecast the stock price. In order to overcome the limitations of traditional prediction of econometric model, people using artificial neural network, support vector machine and genetic algorithm artificial intelligence method to predict the result of the test shows that the method is better than the econometric model the traditional. This paper presents a method of nonlinear combination of EMD and BP neural network prediction based on the method of using EMD technology to stock price series is decomposed into several different frequency components, these components, then we construct neural network model for the prediction of different, the final value of each sequence, then each prediction values are integrated to get the final prediction value.
The empirical study shows that, firstly, the macro economic development China and the stock market is basically the same, the stock price index to a certain extent can reflect the overall level and development trend of China's macro economy. The growth rate of industrial added value, the money growth rate has a positive impact on the stock price, inflation rate and exchange rate, and the interest rate has a negative impact on the stock price, the conclusion and theory are basically the same. In addition, the artificial neural network model to predict the stock price prediction method proposed in this paper can better than the single price.

【学位授予单位】：东北财经大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F224;F832.51

【相似文献】