基于GA-BP神经网络的基金评级研究  

Research on Fund Rating based on GA-BP Neural Network

作　　者：俞军 高振坤 YU Jun;GAO Zhenkun(School of Economics and Management,Hefei University,Hefei 230601,China)

机构地区：[1]合肥大学经济与管理学院,安徽合肥230601

出　　处：《吉林化工学院学报》2023年第5期72-78,共7页Journal of Jilin Institute of Chemical Technology

摘　　要：在中国基金市场上,业绩比较基准以及投资风格多样使得投资者难以鉴别基金业绩的优劣。基金评级能够映射市场信息协助投资者和监管者进行决策,具有重要的参考价值和监督价值。文章采用基金经典评价指标建立一个基金评级指标体系,利用BP神经网络来对基金进行评级,运用具备优秀非线性寻优能力的遗传算法优化BP神经网络的初始权阈值,构造出多输入的GA-BP神经网络基金评级模型。实验证明,经遗传算法优化后的BP神经网络在训练时可以更快地收敛,在仿真能力和误差水平等方面都优于BP神经网络,能够更加准确地评估基金等级。In the Chinese fund market,performance benchmarks and various investment styles make it difficult for investors to identify the merits of fund performance.Fund ratings can map market information to assist investors and regulators in making decisions,and have important reference and monitoring values.This paper used the classic evaluation index of funds to establish a fund rating index system,used the BP neural network to rate the fund,and used the genetic algorithm with excellent nonlinear optimization ability to optimize the initial weight threshold of the BP neural network,and constructed a multi-input GA-BP neural network fund rating model.Experiments showed that the BP neural network optimized by the genetic algorithm could converge faster during training,and was superior to the BP neural network in terms of simulation ability and error level,and could evaluate the fund level more accurately.

关 键 词：BP神经网络 遗传算法 基金评级 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] F830.91[自动化与计算机技术—控制科学与工程]