Reproduce “FinKario: Event-Enhanced Automated Construction of Financial Knowledge Graph” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xiang Li, Penglei Sun, Wanyun Zhou, Zikai Wei, Yongqi Zhang, Xiaowen Chu
Researchers from The Hong Kong University of Science and Technology (Guangzhou) and IDEA developed FinKario, an event-enhanced financial knowledge graph, and FinKario-RAG, a two-stage graph-based retrieval system, to provide dynamic, structured insights for investment analysis. The system achieved an Annualized Rate of Return of 2.633 and a Sharpe Ratio of 4.926 in backtesting, surpassing existing financial LLMs by an average of 18.81% in predictive accuracy.
Key insights
- Dynamically updated event information, captured in the Event graph, is crucial for financial analysis, with its removal leading to an 87.2% drop in Annualized Rate of Return.
- Guiding LLM-driven schema generation with professional financial templates yields high-precision, domain-aligned, and interpretable knowledge graphs.
- A two-stage, graph-based retrieval strategy provides superior semantic coherence and holistic context for LLMs compared to conventional RAG methods.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.