Reproduce “FinCARDS: Card-Based Analyst Reranking for Financial Document Question Answering” 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
Yixi Zhou, Fan Zhang, Yu Chen, Haipeng Zhang, Preslav Nakov, Zhuohan Xie
FinCARDS ranks evidence passages within a single SEC filing for financial questions. It converts passages and questions into finance-specific cards and intents, then uses staged filtering and bootstrap-stable listwise reranking. On FinAgentBench’s intra-document chunk-ranking task, GPT-5 Mini’s nDCG@10 rises from 63.66 after Stage 2 to 71.58 after Stage 3; this measures retrieval, not answer correctness.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.