基于改进小波去噪的支持向量机的股票型基金净值预测研究

发布时间：2018-01-23 22:32

本文关键词： 小波去噪支持向量机金融核基金净值　出处：《重庆师范大学》2014年硕士论文　论文类型：学位论文

【摘要】：最近20年来，投资基金在全球金融市场中以令人瞩目的速度迅猛发展。我们通过基金净值可以看到基金的收益情况，并把它作为是否买基金以及买哪只基金的主要参考依据。因此，对基金净值的预测具有非常重要的应用价值。支持向量机是基于统计学习理论的一种机器学习的方法。作为结果风险最小化准则的具体实现，支持向量机方法具有全局最优，结构简单，推广能力强等优点，已成为近年来机器学习领域最有影响的成果之一。然而现实世界中的数据，通常具有数量大，，不完整，带有噪音等特点。这些带噪音的数据和数据的不完整性将会影响模型预测的准确性。在进行基金净值的预测中，我们最关心的是蕴藏在每日基金净值中的净值的涨跌趋势，而不是那些人为的或者是市场波动造成的每日基金净值的噪声或者毛刺。针对上述两个问题，本文重点研究了基于小波去噪的支持向量机的基金净值预测模型。本文主要工作包括以下几个方面：（1）主要是利用小波去噪方法对数据进行预处理，提出改进的小波阈值去噪方法，并借助Matlab工具进行实证研究，表明改进的小波阈值去噪方法更优。（2）主要是对支持向量机的几种典型的核函数的参数选择进行了详细的研究，并在多项式核和高斯径向基核的基础上，构造了新的核函数：金融核。并借助LibSVM工具包进行实证研究，表明新的核函数具有更好的优越性。（3）建立基于改进小波去噪的支持向量机的股票型基金净值预测模型，对两支基金进行实证分析，并与BP神经网络预测进行比较，表明建立的模型具有很好的预测效果。
[Abstract]:In the past 20 years, investment funds have developed rapidly in the global financial markets. We can see the return of funds through the net value of funds. It is regarded as the main reference of whether to buy the fund and which fund to buy. Therefore, the forecast of the net value of the fund has very important application value. Support vector machine (SVM) is a machine learning method based on statistical learning theory. As a concrete implementation of the result risk minimization criterion, support vector machine (SVM) is globally optimal and simple in structure. In recent years, it has become one of the most influential achievements in the field of machine learning. In the real world, however, data are often large and incomplete. Features such as noise. The incompleteness of these noisy data and data will affect the accuracy of the model forecast. What concerns us most is the upward and downward trend of net worth contained in the net daily fund, not the noise or burr caused by artificial or market volatility. Aiming at the above two problems, this paper focuses on the prediction model of fund net value based on wavelet denoising. The main work of this paper includes the following aspects: 1) the wavelet denoising method is mainly used to pre-process the data, and an improved wavelet threshold de-noising method is proposed, and an empirical study is carried out with the help of Matlab tool. It shows that the improved wavelet threshold denoising method is better. The parameter selection of several typical kernel functions of support vector machine is studied in detail, and based on polynomial kernel and Gao Si radial basis kernel. A new kernel function, financial kernel, is constructed, and an empirical study is carried out with the help of LibSVM Toolkit, which shows that the new kernel function has better advantages. 3) establish the forecasting model of equity fund net value based on improved wavelet denoising support vector machine, and analyze the two funds empirically, and compare them with BP neural network forecast. It shows that the model has good prediction effect.
【学位授予单位】：重庆师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F830.91;TP18

【参考文献】