Rating
1870
Battle Count: 50
Relevance
10/10
Highly relevant as it directly addresses the core challenge of automated factor discovery and provides a rigorous framework for evaluating whether new mining techniques translate into real-world portfolio performance.
Implementation Complexity
8/10
High complexity due to the integration of heterogeneous mining systems (GP, RL, LLM agents), the need for a shared execution contract for symbolic and code-based factors, and the multi-level evaluation pipeline.
Reproducibility
5/5
The paper provides a public GitHub repository with code, data processing scripts, and detailed implementation configurations for all evaluated methods. It uses standard public financial data (Yahoo Finance) and open-source models (Qwen3.8-27B).
About this paper
Methodology: FactorBench. Problem types: Portfolio Optimization, Time Series Forecasting, Ranking, Factor Mining.
The interactive Everscope explorer (charts, battles, favorites) loads below.