基于时间价值的神经网络的股票价格预测

发布时间：2018-01-05 03:35

本文关键词：基于时间价值的神经网络的股票价格预测　出处：《广东财经大学》2013年硕士论文　论文类型：学位论文

【摘要】：股票属于一种高风险、高收益的投资，已成为现代生活中不可缺少的一部分，因此投资者们时刻关心股市，分析股市，研究价格趋势。股票市场中随机因素很多，导致股票价格波动表现出很强的不确定性，以致传统预测技术的效果并不理想。于是，建立一个合理的股票价格预测模型，具有重要的理论意义和实践价值。本文通过深入分析股票原理，比较常见的股票预测方法，探讨BP神经网络在股票预测上的可行性。从原理上讲，神经网络是对股票交易的历史数据学习后实现对未来股票价格的预测。具体而言，BP网络通过对股票的历史数据的学习,不断地修正相应的权值、阀值，最终建立一个相对合理的模型。本文研究的是投机的超短线股票交易，与传统的投资理念有明显区别；预测的结果是来源于多次预测结果的分析，而非特指某一次预测。本文提出了一种创新的研究思想——引入基于时间价值的动态权重误差函数，设计出一种基于时间价值的神经网络模型。本文认为：BP模型通过引入动态权重的方法，可以改变了原来BP模型单纯的拟合训练集数据，更灵活地择优而达到预测效果。据此，本文采用MATLAB软件选定医药行业的股票进行仿真实验。实证结果表明:与传统预测方法和BP神经网络相比，本文提出的模型准确率较高，明显降低预测误差，进一步提高了网络的泛化能力和模型预测精度，优化了股票价格预测效果。为了验证模型的经济和社会效益，，本文设计了一种现实中可实现的模拟交易操作（T+0模型），验证了基于时间价值的BP模型的价值。
[Abstract]:Stock is a kind of high risk, high yield investment, has become an indispensable part of modern life, so investors always care about the stock market, analysis of the stock market. Research on price trend. There are many random factors in stock market, which lead to strong uncertainty of stock price fluctuation, so the effect of traditional forecasting technology is not ideal. It is of great theoretical and practical value to establish a reasonable forecasting model of stock price. This paper discusses the feasibility of BP neural network in stock forecasting by deeply analyzing the stock principle and comparing the common stock forecasting methods. Neural network is to realize the prediction of the future stock price after learning the historical data of stock trading. Specifically, the BP network constantly modifies the corresponding weights and thresholds by learning the historical data of the stock. Finally, a relatively reasonable model is established. This paper studies the speculative ultra-short term stock trading, which is obviously different from the traditional investment concept. The result of prediction comes from the analysis of multiple prediction results, not from a particular prediction. In this paper, an innovative research idea is proposed, which is to introduce the dynamic weight error function based on time value. A neural network model based on time value is designed. This paper holds that the original BP model can change the original BP model by introducing the dynamic weight method. According to this, the MATLAB software is used to select the stocks of the pharmaceutical industry for simulation experiment. The empirical results show that: compared with the traditional forecasting method and BP neural network. The model presented in this paper has a high accuracy, obviously reduces the prediction error, further improves the generalization ability of the network and the prediction accuracy of the model, and optimizes the forecasting effect of stock price, in order to verify the economic and social benefits of the model. In this paper, a realistic simulation transaction operation model is designed, which verifies the value of BP model based on time value.
【学位授予单位】：广东财经大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP183;F830.91

【参考文献】