基于时变概率密度函数的收益率方向预测

发布时间：2018-01-28 21:57

本文关键词： 动态核密度估计时变的股票收益率的方向行动阈值　出处：《浙江工商大学》2017年硕士论文　论文类型：学位论文

【摘要】：在现代整个金融行业不断发展的推动下,股票市场以及证券业在我国越来越受到众多投资者的关注,越来越多的人愿意加入到股票市场中来,投资者们的投资行为以及他们所期望得到丰厚的收益,这在极大程度上促进了我国股票市场的繁荣。而在他们这种投资行为的背后,众多的投资者开始意识到股票市场收益率方向预测的重要性。因此,对股票市场收益率方向变化的深度分析以及预测便有了极大的经济意义以及实际应用的价值。国内外的学者对于股票市场的研究,现也达到了一定的水平。随着信息科技技术的不断发展,新的技术分析和理论知识都不断被注入到数学模型中。由于我国的金融行业逐步规范化,再加上国内外学者和投资者们对于详细分析股票市场的方向变化的迫切需求,都是本文得以发展的源动力。本文研究的主要目的是对股票市场上的收益率进行方向预测,认为股票收益率的方向在一定程度上是可预测的。股票收益率方向预测是指:基于股票市场的历史数据以及其以往的历史趋势来看,在未来一段时间内对股票市场的收益率方向进行预测。本文基于Harvey和Oryshchenko(2012)提出的时变概率密度函数理论的应用和扩展,利用非参数模型对股票市场的收益率进行了方向预测。非参数模型的优势在于其数据不需要满足某些特定的分布。由于本文的样本数据是股票收益率数据,属于时间序列数据,而时间序列数据会受到很多的偶然因素影响,从而表现出随机性。本文在股票市场的历史数据信息集中,用基于时变概率密度函数的非参数模型预测了下一期股票收益率方向的概率,并且利用方向预测和二阶偏上矩之间的关系,提出了方向预测概率的调整机制,也就是本文的创新点。最后还引入了二值选择模型作为对比模型。中国股票市场数据的实证研究表明,我们基于时变概率密度函数的收益率方向预测基准模型和调整机制无论在统计意义上还是在经济意义上,对股票市场收益率方向都有显著的样本外预测的能力。不仅如此,调整机制还显示出了比收益率方向预测基准模型更好的预测能力。最后相比于二值选择模型,无论是收益率方向预测基准模型还是其调整机制的预测效果都优于二值选择模型。
[Abstract]:With the continuous development of the modern financial industry, the stock market and securities industry in China are more and more concerned by many investors, more and more people are willing to join the stock market. The investment behavior of investors and the huge profits they expect have greatly promoted the prosperity of the stock market in our country. But behind this kind of investment behavior. Many investors are beginning to realize the importance of the stock market yield direction forecast. It has great economic significance and practical application value to analyze and forecast the change of return direction of stock market. The scholars at home and abroad have studied the stock market. With the continuous development of information technology, new technical analysis and theoretical knowledge have been continuously injected into the mathematical model. Due to the gradual standardization of the financial industry in China. Coupled with the domestic and foreign scholars and investors for detailed analysis of the direction of the stock market changes in the urgent need. The main purpose of this paper is to forecast the return rate in the stock market. It is believed that the direction of stock return is predictable to a certain extent. The direction prediction of stock return refers to: based on the historical data of stock market and its historical trend. This paper is based on Harvey and Oryshchenko 2012). The application and extension of time-varying probability density function theory proposed in this paper. The non-parametric model is used to predict the return rate of the stock market. The advantage of the non-parametric model is that its data do not need to satisfy some specific distribution. Because the sample data in this paper is the stock yield data. Time series data, and time series data will be affected by a lot of accidental factors, thus showing randomness. This paper in the stock market historical data set. The non-parametric model based on time-varying probability density function is used to predict the probability of the next stock return direction, and the adjustment mechanism of the direction prediction probability is proposed by using the relation between the direction prediction and the second order upper-moment. This is the innovation of this paper. Finally, the binary selection model is introduced as the contrast model. The empirical study of Chinese stock market data shows that. We based on the time-varying probability density function of the return direction prediction benchmark model and adjustment mechanism in the statistical sense or in the economic sense. Moreover, the adjustment mechanism also shows a better prediction ability than the return direction prediction benchmark model. Finally, compared with the binary selection model. Both the return direction prediction benchmark model and its adjustment mechanism are superior to the binary selection model.
【学位授予单位】：浙江工商大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F832.51

【相似文献】