基于深度学习的收益率预测与投资组合模型  

Deep learning-based yield rate prediction and portfolio model

作　　者：张博群 沈虹[1] ZHANG Boqun;SHEN Hong(School of Business,Yangzhou University,Yangzhou Jiangsu 225127,China)

机构地区：[1]扬州大学商学院,江苏扬州225000

出　　处：《阜阳师范大学学报(自然科学版)》2024年第4期91-97,共7页Journal of Fuyang Normal University:Natural Science

基　　金：国家自然科学基金资助项目(61803331,92371116);江苏省自然科学基金资助项目(BK20170515)。

摘　　要：在资产池中选出具有高回报的优质资产,是投资组合模型的关键因素。文章采用长短期记忆网络这一深度学习算法对资产收益进行预估,提高了精准识别高质量资产的能力,并因此增进投资组合策略的表现;然后使用均值-方差模型对选出的优质资产进行投资比例划分;最后以中国大宗商品期货市场为资产池,进行100期投资,从收益率、风险和风险收益三个方面选取不同指标来评估投资组合的绩效,结果表明LSTM+MV投资组合在风险绩效上表现最优,且取得较好的收益。Identifying high-return, high-quality assets within an asset pool is a critical component of portfolio modeling.This study employs the Long Short-Term Memory(LSTM) network, a deep learning algorithm, to forecast asset returns, thereby enhancing the precision of high-quality asset identification and subsequently improving the overall performance of the portfolio strategy. Following asset selection, the Mean-Variance(MV) model is applied to allocate investment proportions among the identified high-quality assets. Using China's commodity futures market as the asset pool, a 100-period investment simulation is conducted. Portfolio performance is evaluated through various metrics encompassing return, risk, and risk-adjusted return. The results demonstrate that the LSTM+MV portfolio model achieves superior performance in risk management and yields favorable returns, showcasing optimal efficacy in risk-adjusted performance.

关 键 词：投资组合 深度学习 均值-方差模型 

分 类 号：F7323.7[经济管理—产业经济] TP391.9[自动化与计算机技术—计算机应用技术]