基于神经网络的股票价格走势预测及其MATLAB实现 - 论文

基于神经网络的股票价格走势预测及其MATLAB实现

摘要

伴随着我国经济的高速发展和广大投资者日益旺盛的需求，股票投资已经成为一种常见的投资手段，而股票价格预测也逐渐成为广大投资者关心和研究的重点问题。股票价格的波动是一个高度复杂化的非线性动态系统，其本身具有诸如大规模数据、噪声、模糊非线性等特点。针对这些特点本文在深入分析股票市场实际预测中所面临的关键问题和比较各种已有的股票预测方法的基础上，探讨运用神经网络这一人工智能工具，研究基于历史数据分析的股票预测模型。

神经网络是建立在对大规模的股票历史数据的学习仿真的基础上，运用黑盒预测方式找出股市波动的内在规律，并通过将其存储在网络的权值、阈值中，以此来预测未来短期或是中长期的价格走势。

关键字：神经网络，股票，预测，MATLAB工具箱

ABSTRACT

Along with the economy growth and increasingly strong demand of many investors in our country, stock has become a common means of investment, and stock price forecast has greatly been one of the focuses of study topic. The change of stock price is a highly complicated nonlinear dynamic system, itself has many characteristics such as massive data, noise, fuzzy and nonlinear. This article analyses the key issues being existent in the real stock market prediction and compares various existing stock forecasting methods. We will try to research on stock price prediction model based on a neural network with huge historical data.

Neural network is based on studying massive historical data, uses the black box of forecasting ways to find the internal disciplinarian of stock market, and stores them in the weights and valves values of the neural network for predicting the short-term or long-term trend in the future.

KEYWORD：Neural networks, Stock, prediction, MATLAB toolbox

1

目录

摘要 .................................................................................................................................................. 1 一． 绪论......................................................................................................................................... 3

1.1研究背景及意义 ................................................................................................................. 3 1.2国内外研究的现状 ............................................................................................................. 4 1.3 论文的研究方法及其框架结构 ........................................................................................ 5 二． 股票预测的关键问题分析 ..................................................................................................... 6

2.1 股票价格波动的因素以及预测的基本假设 .................................................................... 6 2.2 股票预测的常用术语和技术指标 .................................................................................... 8 2.3股票数据的特点 ............................................................................................................... 11 三． 神经网络的基本原理介绍 ................................................................................................... 12

3.1人工神经网络的定义和发展过程 ................................................................................... 12 3.2 神经网络基本原理 .......................................................................................................... 12 3.3 BP神经网络介绍 ............................................................................................................. 15 3.4 神经网络的特点 .............................................................................................................. 19 3.5 神经网络的在实际预测模型中的问题 .......................................................................... 20 四． 神经网络算法 ....................................................................................................................... 21

4.1 输出输入变量的选取 ...................................................................................................... 21 4.2数据归一化处理 ............................................................................................................... 22 4.3数据样本分类 ................................................................................................................... 22 4.4网络初始化 ....................................................................................................................... 22 4.5 训练网络.......................................................................................................................... 23 4.6网络仿真........................................................................................................................... 24 五． 仿真实验 ............................................................................................................................... 24

4.1 单日收盘价对单日收盘价预测 ...................................................................................... 24 4.2 单日收盘价，成交量对单日收盘价预测 ...................................................................... 25 4.3 多日收盘价，成交量对单日收盘价预测 ...................................................................... 26 4.4 多日收盘价，成交量对多日收盘价预测 ...................................................................... 27 六． 结论和展望 ........................................................................................... 错误！未定义书签。 附录（股票数据和程序代码） ..................................................................................................... 28

2

一． 绪论

1.1研究背景及意义

许多伟大的预言家认为，每一个世纪的开始十年发生的事情将奠定这个世纪的发展方向。进入新的21世纪以来，头十年发生的最重要的一件事情莫过于2008年的金融危机。这场罕见的金融风暴不仅给中国带来了巨大的冲击，也给国际市场带来了更多的挑 …… 此处隐藏：8037字，全部文档内容请下载后查看。喜欢就下载吧 ……