20230913-东证期货-基本面量化专题报告_从因果涌现的视角挖掘择时模型_35页_4mb

Summary of Causal Emergence Approach for Timining Models

This report analyzes the application of causal emergence theory to build择时 (timing) models for financial assets. The methodology involves transforming time series data through coarse-graining to create network graphs, computing the effective information (EI) metric, and comparing it with traditional network topology measures like average shortest path length. The analysis demonstrates that macro-scale causal properties provide superior predictive power for asset timing, offering a new framework for quantitative modeling in complex financial systems.

Key Findings from Empirical Testing:

• The EI-based models generated positive returns across multiple assets, with maximum win rates exceeding 53% for indices like Shanghai Composite and USD Index.
• Compared to average shortest path length models, EI showed superior performance in capturing market trends, particularly for slower-moving assets like bonds and gold.
• Parameters such as coarse-graining length and slope factors were optimized for each asset, with notable differences in performance outcomes.

Conclusion:
The study confirms that causal emergence can enhance timing strategies by focusing on system-wide causality. Future work should explore further refinements and applications of complex systems theory in finance.