摘要
针对股票收益率和风险的不确定性,文章提出一种基于BP神经网络的马尔科夫链和遗传算法组合模型,该模型通过对神经网络以滚动预测法完成股票价格曲线的粗略拟合;在此基础上,借助马尔科夫链对股票价格的曲线拟合进行系统状态划分,并用遗传算法对系统状态划分进行优化,提高马尔科夫系统状态划分的合理性;最后用马尔科夫链缩小预测区间以提高预测精确度。
As the stock return and the risk are quite unpredictable, the paper proposes a prediction model that combines Markov chain and genetic algorithm based on BP neural network. The model fits roughly the real stock price curve by means of rolling forecast through the neural network, on the basis of which, Markov chain is used to achieve the system state division of the curve fitting of stock price and then genetic algorithm is employed to optimize the division in order to improve its accuracy before Markov chain is applied again to reduce the prediction interval for more accuracy.
出处
《宁波工程学院学报》
2012年第3期29-37,共9页
Journal of Ningbo University of Technology
关键词
BP神经网络
马尔科夫链
遗传算法
投资组合
BP neural network
Markov chain
genetic algorithm
investment portfolio
