Momentum-integrated Multi-task Stock Recommendation with Converge-based Optimization

2025/08/05 by Hao Wang, Wang, Hao, Jingshu Peng +11 · 1 voice
Business, Management and Accounting · Computer Science · Decision Sciences · Economics, Econometrics and Finance · #Customer churn and segmentation #Stock Market Forecasting Methods #cs.AI #cs.LG #q-fin.ST

paper · pdf · doi:10.48550/arxiv.2509.10461

arxiv published 2025/08/05 · arxiv updated 2026/01/24

Abstract

Stock recommendation is critical in Fintech applications, which leverage price series and alternative information to estimate future stock performance. Traditional time-series forecasting training often fails to capture stock trends and rankings simultaneously, which are essential factors for investors. To tackle this issue, we introduce a Multi-Task Learning (MTL) framework for stock recommendation, Momentum-integrated Multi-task Stock Recommendation with Converge-based Optimization (MiM-StocR). To improve the model's ability to capture short-term trends, we incorporate a momentum line indicator in model training. To prioritize top-performing stocks and optimize investment allocation, we propose a listwise ranking loss function called Adaptive-k ApproxNDCG. Moreover, due to the volatility and uncertainty of the stock market, existing MTL frameworks face overfitting issues when applied to stock time series. To mitigate this issue, we introduce the Converge-based Quad-Balancing (CQB) method. We conducted extensive experiments on three stock benchmarks: SEE50, CSI 100, and CSI 300. MiM-StocR outperforms state-of-the-art MTL baselines across both ranking and profitability evaluations.

Citations

• On Evaluating Loss Functions for Stock Ranking: An Empirical Analysis With Transformer Model
• DeepUnifiedMom: Unified Time-series Momentum Portfolio Construction via Multi-Task Learning with Multi-Gate Mixture of Experts
• Pairwise Ranking Loss for Multi-Task Learning in Recommender Systems
• Dual-Balancing for Multi-Task Learning
• Methods for Acquiring and Incorporating Knowledge into Stock Price Prediction: A Survey
• DoubleAdapt: A Meta-learning Approach to Incremental Learning for Stock Trend Forecasting
• Multi-Task Learning as a Bargaining Game
• NumHTML: Numeric-Oriented Hierarchical Transformer Model for Multi-task Financial Forecasting
• HIST: A Graph-based Framework for Stock Trend Forecasting via Mining Concept-Oriented Shared Information
• Conflict-Averse Gradient Descent for Multi-task Learning
• Learning Multiple Stock Trading Patterns with Temporal Routing Adaptor and Optimal Transport
• Constructing long-short stock portfolio with a new listwise\n learn-to-rank algorithm
• Qlib: An AI-oriented Quantitative Investment Platform
• Gradient Surgery for Multi-Task Learning
• Multi-Task Learning as Multi-Objective Optimization
• Temporal Relational Ranking for Stock Prediction
• Listening to Chaotic Whispers: A Deep Learning Framework for News-oriented Stock Trend Prediction
• Graph Attention Networks
• A Survey on Multi-Task Learning
• A Survey on Multi-Task Learning
• Depth Map Prediction from a Single Image using a Multi-Scale Deep Network
• A Theoretical Analysis of NDCG Type Ranking Measures
• Long Short-Term Memory
• Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency
• Common risk factors in the returns on stocks and bonds

Discussions

• Hao Wang, Jingshu Peng, Yanyan Shen, Xujia Li, Lei Chen: Momentum-integrated Multi-task Stock Recommendation with Converge-based Optimization https://arxiv.org/abs/2509.10461 https://arxiv.org/pdf/250 [bsky, 0 points, 0 comments]

Related

• A Practical Machine Learning Approach for Dynamic Stock Recommendation
• From Headlines to Holdings: Deep Learning for Smarter Portfolio Decisions
• Supervised classification-based stock prediction and portfolio optimization
• Multi-task Learning for Financial Forecasting
• Deep Learning for Short Term Equity Trend Forecasting: A Behavior Driven Multi Factor Approach