基于投资者情绪和NARX动态神经网络的证券价格可预测性研究

发布时间：2018-04-04 07:10

本文选题：投资者情绪　切入点：动态神经网络　出处：《首都经济贸易大学》2017年硕士论文

【摘要】：金融证券市场上充满着各种不确定性,现代金融理论正是建立在这种随机不确定性的假设之上。证券市场瞬息万变,是一个动态的、非线性的、充满模糊性的复杂系统,而传统的线性时间序列预测模型并不能很好的适应金融市场的这种模糊不确定性。近年来,随着计算机技术和行为金融学的发展,基于心理学的投资者行为研究、基于人工智能技术的人工神经网络研究如火如荼的发展起来,并在理论及实践中展现了其独特魅力。本文的研究中,首先介绍了行为金融学的投资者情绪这种反映投资者在证券市场交易中的非理性因素,并构建了能够在一定程度上衡量投资者情绪的情绪指数。构建投资者情绪综合指数时,本文选择了A股市场换手率(TURN)、A股成交量(VOL),消费者信心指数(CCI)、A股股票市场活跃账户数(ASA)和百度指数(BDI)五个变量,作为投资者情绪的代理指标,并参照Baker和Wurgle构建BW情绪指数的方法-主成分分析方法,构建了投资者情绪综合指数SENT。然后,本文引入了NARX动态神经网络预测模型,创造性地将投资者情绪加入至神经网络模型中,作为神经网络的一个输入,从而构建了基于投资者情绪的NARX动态神经网络预测模型。将投资者情绪这种非理性因素和神经网络这种具有较强适应性学习能力的模型相结合,运用于对证券市场价格的预测中,实证结果表明,基于投资者情绪的动态神经网络对证券价格的预测准确性优于传统的线性ARIMA模型。
[Abstract]:The financial securities market is full of various uncertainties. Modern financial theory is based on the assumption of random uncertainty.The securities market is a dynamic, nonlinear and fuzzy complex system, but the traditional linear time series prediction model can not adapt to the fuzzy uncertainty of the financial market.In recent years, with the development of computer technology and behavioral finance, the research of investor behavior based on psychology and artificial neural network based on artificial intelligence technology is in full swing.And in theory and practice to show its unique charm.In this paper, we first introduce the investor sentiment of behavioral finance, which reflects the irrational factor of investor trading in the stock market, and construct the emotional index which can measure investor sentiment to a certain extent.In order to construct the composite index of investor sentiment, this paper chooses five variables as proxy index of investor sentiment: turnover rate of A-share market, turnover rate of A-share market, consumer confidence index (CCI), active account number of A-share stock market (ASAs) and Baidu index (BDI).According to the method of Baker and Wurgle to construct BW emotion index, the principal component analysis (PCA) method is used to construct investor sentiment index (set).Then, the NARX dynamic neural network prediction model is introduced, which creatively adds investor sentiment to the neural network model as an input of the neural network.Thus, the prediction model of NARX dynamic neural network based on investor sentiment is constructed.This paper combines the irrational factor of investor sentiment with the model of neural network, which has strong adaptive learning ability, and applies it to the prediction of stock market price. The empirical results show that,The prediction accuracy of dynamic neural network based on investor sentiment is better than that of traditional linear ARIMA model.
【学位授予单位】：首都经济贸易大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP183;F832.51

【相似文献】