基于长短期记忆神经网络的金融压力指数预测  

The Prediction of Financial Pressure Index Based on Long-term and Short-term Memory Neural Network

作　　者：张永[1] 郑锋淇 杨兴雨[1] 赵雪瑾 ZHANG Yong;ZHENG Feng-qi;YANG Xing-yu;ZHAO Xue-jin(School of Management,Guangdong University of Technology,Guangzhou 510520,China;School of Economics,Guangdong University of Technology,Guangzhou 510520,China)

机构地区：[1]广东工业大学管理学院,广东广州510520 [2]广东工业大学经济学院,广东广州510520

出　　处：《系统工程》2023年第5期115-123,共9页Systems Engineering

基　　金：教育部人文社会科学研究基金资助项目(21YJA630117);国家自然科学基金资助项目(72371080);广东省自然科学基金资助项目(2023A1515012840);广东省哲学社会科学规划项目(GD19CGL06,GD23XGL022)。

摘　　要：长短期记忆神经网络是一种深度学习方法,能够用于挖掘金融时间序列历史数据的信息和研究序列依赖关系。本文首先从货币、债券、股票、外汇、银行5个子市场选取10个市场指标构建金融压力指数;然后把当期和上期的金融压力指数作为输入特征,应用长短期记忆神经网络预测金融压力指数,并将其与自回归滑动平均模型的预测结果进行对比。结果表明,长短期记忆神经网络具有更小的预测误差。此外,预测结果显示,我国未来短期内的金融压力总体呈现稳定趋势,整体上看风险是可控的。Long-short memory neural network is a deep learning method that can be used to mine information from historical data of financial time series and to study sequence dependence.The paper first selects 10 market indicators from five submarkets:currency,bond,stock,foreign exchange,and bank to construct financial stress index;and then takes the current and previous financial stress indices as input features,applies the long-term and short-term memory neural network to predict financial stress indices,and compares it with the auto-regressive and moving average model.The results show that the long-and short-term memory neural network has a smaller lower prediction error.In addition,the prediction results show that China's financial pressure will generally shows an stable trend in the near future and the overall risk is controllable.

关 键 词：金融压力指数 预测 深度学习 长短期记忆神经网络 

分 类 号：F832[经济管理—金融学]