Combined machine learning for stock selection strategy based on dynamic weighting methods

2025-08-25

AI Summary

This paper addresses two key limitations in stock selection: weak generalization of single models and poor adaptability due to static weighting schemes. To overcome these, we propose an ensemble learning framework that dynamically weights constituent models based on the Information Coefficient (IC). Specifically, we integrate three representative models—Random Forest, XGBoost, and LSTM—and design a rolling-IC-driven real-time weight allocation mechanism. Additionally, a factor importance screening module is embedded to optimize input features. Backtesting on CSI 300 constituents demonstrates that our framework outperforms both conventional static-weight ensembles and individual models: it achieves a 12.3% higher annualized return and attains a mean IC of 0.048 (p < 0.01), indicating significantly improved prediction stability and alpha generation capability. The core contributions are: (i) elevating IC from a mere performance evaluation metric to a principled basis for dynamic weight generation; and (ii) empirically validating that factor screening critically enhances ensemble effectiveness.

Technology Category

• Machine Learning: Ensemble Methods
• Reasoning under Uncertainty: Stochastic Optimization
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

This paper proposes a novel stock selection strategy framework based on combined machine learning algorithms. Two types of weighting methods for three representative machine learning algorithms are developed to predict the returns of the stock selection strategy. One is static weighting based on model evaluation metrics, the other is dynamic weighting based on Information Coefficients (IC). Using CSI 300 index data, we empirically evaluate the strategy' s backtested performance and model predictive accuracy. The main results are as follows: (1) The strategy by combined machine learning algorithms significantly outperforms single-model approaches in backtested returns. (2) IC-based weighting (particularly IC_Mean) demonstrates greater competitiveness than evaluation-metric-based weighting in both backtested returns and predictive performance. (3) Factor screening substantially enhances the performance of combined machine learning strategies.

Problem

• Combining machine learning algorithms for stock selection strategy
• Developing dynamic weighting methods to predict stock returns
• Evaluating strategy performance using CSI 300 index data

Innovation

• Combined machine learning algorithms for stock selection
• Dynamic weighting using Information Coefficients (IC)
• Factor screening enhances strategy performance