数据分布可变的股市拐点预测方法  

Stock turning point prediction with variable data distribution

作　　者：姚宏亮[1] 施振振 王浩[1] YAO Hongliang;SHI Zhenzhen;WANG Hao(School of Computer and Information,Hefei University of Technology,Hefei 230601,China)

机构地区：[1]合肥工业大学计算机与信息学院

出　　处：《合肥工业大学学报（自然科学版）》2020年第1期45-51,共7页Journal of Hefei University of Technology：Natural Science

基　　金：国家重点基础研究发展计划(973计划)资助项目(2013CB329604);国家自然科学基金资助项目(61175051;61175033)

摘　　要：文章研究了股市趋势预测中的拐点发现问题,提出一种融合能量特征的支持向量机(support vector machine,SVM)股市拐点预测方法。该方法在原加权支持向量机(weighted support vector machine,W-SVM)模型中引入平衡项得到平衡加权支持向量机(balanced W-SVM,BW-SVM)模型,新模型对历史数据分布和近期局部数据分布进行了融合,平衡项是用历史数据训练所得参数和用当前局部数据训练所得参数的均方误差,利用调节参数μ对均方误差项进行调节。文章在能体现数据分布变化技术指标的基础上,进一步量化提取拐点的能量特征,并利用马尔可夫毯融合拐点的能量特征,最后将能量特征带入BW-SVM模型,提出基于能量的平衡加权支持向量机(BW-SVM based on energy,EBW-SVM)算法对股市拐点进行预测。实验结果表明,EBW-SVM算法具有良好的性能。This paper studies the problem of finding turning point in stock market trend forecasting,presenting a prediction method for the stock market turning point of support vector machine(SVM)with the characteristics of fusion energy.This method introduces a balance term into the original weighted support vector machine(W-SVM)model leading to a balanced W-SVM(BW-SVM)model.The new model combines historical data distribution and recent local data distribution.The balance term is the mean square error of the parameters obtained from historical data and the current local data,and the mean square error is adjusted by adjustment parameterμ.On the basis of the technical indicators that can reflect the data distribution and change,the energy characteristics of the turning point are further extracted,and Markov blanket is used to fuse the energy characteristics of the turning point.Finally,the energy characteristics are introduced into the BW-SVM model to get BW-SVM based on energy(EBW-SVM)to predict the turning point of the stock market.The experiment shows that the EBW-SVM model has better performance than the traditional method.

关 键 词：拐点预测 支持向量机 数据分布 马尔科夫毯 能量计算 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]