基于多模态特征小波分解的深度学习股价概率预测

doi:10.11896/jsjkx.240600140

摘要/Abstract

摘要： 构建了一种创新的基于多模态特征小波分解的深度学习股价概率预测模型(MWDPF)。该模型融合了动态连续特征、动态分类特征、静态连续特征和静态分类特征等多源异构信息,通过并行融合的策略充分挖掘不同特征子空间的互补信息,全面刻画影响股价波动的多重维度。该模型采用自回归递归神经网络架构,能够直接输出股价变化的概率分布预测,而非单一确定值预测,更加贴近实际股价呈概率分布的特征。另外,该模型引入小波分解技术,对原始时间序列进行去噪,自适应地过滤掉不同尺度下的噪声成分,提高了对内在波动规律的捕捉能力。实证分析阶段,采集了来自金融数据库和互联网论坛的多模态数据,通过缺失值填充、去极值、时间对齐等一系列预处理,以及精心的特征工程和模型优化,实现了优秀的预测性能,显著优于传统的统计学模型和深度学习模型,评价指标均有大幅改善。该模型产生的预测结果被用于构建了一个多因子选股策略,在实际回测中取得了可观的超额收益,进一步验证了该模型在实际投资决策中的有效性。该研究为股价预测提供了一种行之有效的解决方案,丰富了量化投资的理论和方法,具有重要的理论意义和应用价值。

关键词: 概率密度预测, 多模态异构特征融合, 小波分解时频分析, 自回归递归神经网络, 投资组合超额收益

Abstract: This paper constructs an innovative deep learning model for probabilistic stock price prediction based on multi-modal feature wavelet decomposition(MWDPF).This model integrates multi-source heterogeneous information,including dynamic continuous features,dynamic categorical features,static continuous features,and static categorical features.Through a parallel fusion strategy,it fully explores the complementary information in different feature subspaces,comprehensively characterizing the multiple dimensions affecting stock price fluctuations.It adopts an auto-regressive recurrent neural network architecture,which can directly output the probability distribution prediction of stock price changes,rather than a single deterministic value prediction,more closely matching the actual probabilistic distribution characteristics of stock prices.Additionally,this model introduces wavelet decomposition technology to denoise the original time series,adaptively filtering out noise components at different scales,improving its ability to capture intrinsic fluctuation patterns.In the empirical analysis phase,this study collects multi-modal data from financial databases and internet forums,and through a series of preprocessing steps such as missing value imputation,outlier removal,and time alignment,as well as careful feature engineering and model optimization,achieves excellent prediction perfor-mance,significantly outperforming traditional statistical models and deep learning models,with substantial improvements in eva-luation metrics.The prediction results generated by the proposed model are used to construct a multi-factor stock selection strategy,achieving considerable excess returns in real-world backtesting,further verifying the effectiveness of the model in practical investment decision-making.This study provides an effective solution for stock price prediction,enriches the theories and methods of quantitative investment,and has significant theoretical and application value.

Key words: Probability density prediction, Multi-modal heterogeneous feature fusion, Wavelet decomposition time-frequency ana-lysis, Auto-regressive recurrent neural network, Portfolio excess returns

中图分类号:

F224-39

张永宇, 郭晨娟, 魏涵玥. 基于多模态特征小波分解的深度学习股价概率预测[J]. 计算机科学, 2025, 52(6A): 240600140-11. https://doi.org/10.11896/jsjkx.240600140

ZHANG Yongyu, GUO Chenjuan, WEI Hanyue. Deep Learning Stock Price Probability Prediction Based on Multi-modal Feature Wavelet Decomposition[J]. Computer Science, 2025, 52(6A): 240600140-11. https://doi.org/10.11896/jsjkx.240600140

参考文献

[1]MARKOWITZ H.Portfolio Selection[J].The Journal of Fi-nance,1952,7(1):77-91.
[2]SHARP W F.Capital asset prices:Atheory of market equilibriumunder conditions of risk[J].The Journal of Finance,1964,19(3):425-442.
[3]FAMA E F,FRENCH K R.The Cross-Section of ExpectedStock Return[J].The Journal of Finance,1992,47(2):427.
[4]FAMA E F,FREMCH K R.A five-factor asset pricing model[J].Journal of Financial Economics,2015,116(1):1-2.
[5]WU S N,XU N H.A comparative study on the rational assetpricing model and irrational asset pricing model:Evidence from stock market in China[J].Economic Research Journal,2004(6):105-116.
[6]GAO C T,ZHOU X H.Corporate profitability,investment and asset pricing:An empirical research about Chinese stock market[J].Journal of Industrial Engineering and Engineering Management,2016,30(4):25-33.
[7]LI Z B,YANG G Y,FENG Y C,et al.Fama-French five factor model in China stock market[J].Journal of Financial Research,2017(6):191-206.
[8]MENG X J,MENG X L,HU Y Y.Research on Investor Sentiment Index Based on Text Mining and Baidu Index[J].Macroeconomics,2016(1):144-153.
[9]MENG Y,CHANG J.The Impact of Investor Sentiment on Size Effect[J].Journal of Statistics and Information,2019(4):14.
[10]YOU J X,WU J.Spiral of Silence:Media Sentiment and the Asset Mispricing[J].Economic Research Journal,2012(7):141-152.
[11]DI MATTEO T.Muti-scaling in finance[J].Quantitative Fi-nance,2007,7(1):21-36.
[12]WANG C L,LIU G,WANG Q.Research on Multi-FactorQuantitative Stock Selection Model and Timing Strategy[J].Journal of Dongbei University of Finance and Economics,2018(5):81-87.
[13]LIN Y,GUO H,HU J.An SVM-based approach for stock market trend prediction[C]//The 2013 International Joint Conference on Neural Networks(IJCNN).IEEE,2013:1-7.
[14]CAO J,LI Z,LI J.Financial time series forecasting model based on CEEMDAN and LSTM[J].Physical A:Statistical Mechanics and its Applications,2019,519:127-139.
[15]QIAO R Y.Stock Prediction Model Based on Neural Network[J].Operations Research and Management Science,2019,28(10):132-140.
[16]TONG W M,LI Y J,SHAN Y Z.Wavelet analysis based time series data mining [J].Computer Engineering,2008(1):26-29.
[17]HAN Z M,MENG Y X,GUO H Y,et al.Stock price prediction based on dynamic heterogeneous network[J].Application Research of Computers,2024,41(7):2126-2133.
[18]MENG Z Q,ZHU H Q.Application of variable structure temporal neural network model in stock forecasting[J].Computer Engineering and Design,2020,41(06):1733-1741.
[19]LIU W L,ZHANG Y,YANG X Y.Online margin trading stra-tegy based on LSTM prediction information[J].Systems Engineering-Theory & Practice,2024,44(8):2493-2508.
[20]WANG P P,ZHOU X P,CHEN X Y.Multi-feature Nested Attention Fusion of Stock Price and Emotion for Stock Price Prediction[J/OL].[2024-03-13].http://kns.cnki.net/kcms/detail/43.1115.N.20240119.1539.004.html.
[21]ZHAO M,XIE L,LIN W J,et al.Deep Reinforcement Learning Portfolio Model Based on Dynamic Selectors[J].Computer Science,2024,51(4):344-352.
[22]HUANG H J,LI B.Stock price prediction based on VMD-CSSA-LSTM combination model[J].Journal of Nanjing University of Information Science & Technology,2023(11).
[23]SALINAS D,FLUNKERT V,GASTHAUS J,et al.DeepAR:Probabilistic forecasting with autoregressive recurrent networks[J].International Journal of Forecasting,2020,36(3):1181-1191.
[24]WEI Y T,WANG J K,WANG C R,et al.Network Traffic Prediction Algorithm Based on Wavelet Transform and Combinational Models[J].Journal of Northeastern University(Natural Science),2011(10).
[25]WANG X L,ZHANG W H,WANG H Z.Non-stationary time series prediction based on wavelet analysis and AR-LSSVM[J].Control and Decision,2008(3).
[26]YE L S,HE Z X.Multiscale time series anomaly detection incorporating wavelet decomposition[J].Journal of Computer Applications,2024(10).
[27]HUO Y L.Neural Network Prediction with Wavelet Decomposition:A case on maize commodities price sequence[J].Science Technology and Industry,2023,23(14):259-264.
[28]XU W.Application of wavelet analysis in the field of economy and finance[J].China Circulation Economy,2022(30):149-152.
[29]LI S,LIAO W,CHEN Y,et al.PEN:Prediction-ExplanationNetwork to Forecast Stock Price Movement with Better Explainability[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2023:5187-5194.
[30]OUYANG H B,HUANG K,YAN H J.Prediction of Financial Time Series Based on LSTM Neural Network[J].Chinese Journal of Management Science,2020,28(4):27-35.
[31]ZUO Y F,DAI L,WU L C.Statistical Analysis of Log-Return Series of Shenzhen and Shanghai Stock Market[J].Journal of Kunming University of Science and Technology(Natural Science)),2009,34(3):112-116.
[32]DENG F Y,CEN S Q,ZHONG F Q,et al.Short-Term PriceTrend Forecast Based on LSTM Neural Network[J].Computer Systems & Applications,2021,30(4):187-192.
[33]ZHANG X Y,LI P R.The Informed Trading Anomaly in China’s A-Share Market--Research Based on Ensemble Neural Network[J].China Journal of Econometrics,2023,3(3):683-706.
[34]ZHANG H,SHEN H L,LIU Y C.The Study on Multi-factor Quantitative Stock Selection Based on Self-attention Neural Network[J].Journal of Applied Statistics and Management,2020,39(3):556-570.
[35]YUAN J,PAN S,XIE H,et al.Stock Price Prediction Integrating Investor Sentiment Based on S＿AM＿BiLSTM Model[J].Computer Engineering and Applications,2024,60(7).
[36]WAN S F,TANG Y,LIN J J.Integral Sentiment,Background Sentiment and Stock Market Volatility:An Empirical Study Based on TVP-SV-VAR Model[J].Financial Theory & Practice,2023(12):105-115.
[37]ZHANG J X,ZHAO Y,YANG X J,et al.Big Data Analytics,Analyst Reports on Investment Apps and Investors’ Investment Judgments－An Experimental Research[J].Accounting Research,2023(7):166-177.
[38]SHI Q C,XU L Y.Empirical Research on the Listed Compa-nies’ Stock Prices Affected by Negative Public Opinion[J].Journal of Central University of Finance & Economics,2014(10):54-62.
[39]LI Y X,LIU Y Y,GAO L.Research on Risk Spillover Effects Between China Securities Bank Index and Its Ten Weighted Stocks[J].Mathematics in Practice and Theory,2020,50(24):45-54.
[40]DENG W,TAO Y.Industry Peer Effect in Voluntary Earnings Forecast of China’s Listed Companies[J].Financial Regulation Research,2022(9):41-57.
[41]DENG J L.Research on Phenomenon of “High Price Premium” in A-share Market[J].Journal of Shenzhen Polytechnic,2019,18(6):11-21.
[42]HUANG X W,CHENG M,YIN Q W,et al.Corporate Market Caps and the Effect of Media Attention to the Market[J].Financial Economics Research,2018,33(3):68-79.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed