dorsal/arxiv
View SchemaFinCARDS: Card-Based Analyst Reranking for Financial Document Question Answering
| Authors | Yixi Zhou, Fan Zhang, Yu Chen, Haipeng Zhang, Preslav Nakov, Zhuohan Xie |
|---|---|
| Categories | |
| ArXiv ID | 2601.06992vv1 |
| URL | https://arxiv.org/abs/2601.06992 |
| License | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ |
Abstract
Financial question answering (QA) over long corporate filings requires evidence to satisfy strict constraints on entities, financial metrics, fiscal periods, and numeric values. However, existing LLM-based rerankers primarily optimize semantic relevance, leading to unstable rankings and opaque decisions on long documents. We propose FinCards, a structured reranking framework that reframes financial evidence selection as constraint satisfaction under a finance-aware schema. FinCards represents filing chunks and questions using aligned schema fields (entities, metrics, periods, and numeric spans), enabling deterministic field-level matching. Evidence is selected via a multi-stage tournament reranking with stability-aware aggregation, producing auditable decision traces. Across two corporate filing QA benchmarks, FinCards substantially improves early-rank retrieval over both lexical and LLM-based reranking baselines, while reducing ranking variance, without requiring model fine-tuning or unpredictable inference budgets. Our code is available at https://github.com/XanderZhou2022/FINCARDS.
{
"annotation_id": "a372376d-5f23-4fe2-ad5a-f7db6803ec1c",
"date_created": "2026-02-17T05:53:08.755000Z",
"date_modified": "2026-02-17T05:53:08.755000Z",
"file_hash": "79161f8a56fa2bb836133b7f8a8ab6e4b68c07f6077ba7bf1a8376e025117cfb",
"private": false,
"record": {
"abstract": "Financial question answering (QA) over long corporate filings requires evidence to satisfy strict constraints on entities, financial metrics, fiscal periods, and numeric values. However, existing LLM-based rerankers primarily optimize semantic relevance, leading to unstable rankings and opaque decisions on long documents. We propose FinCards, a structured reranking framework that reframes financial evidence selection as constraint satisfaction under a finance-aware schema. FinCards represents filing chunks and questions using aligned schema fields (entities, metrics, periods, and numeric spans), enabling deterministic field-level matching. Evidence is selected via a multi-stage tournament reranking with stability-aware aggregation, producing auditable decision traces. Across two corporate filing QA benchmarks, FinCards substantially improves early-rank retrieval over both lexical and LLM-based reranking baselines, while reducing ranking variance, without requiring model fine-tuning or unpredictable inference budgets. Our code is available at https://github.com/XanderZhou2022/FINCARDS.",
"arxiv_id": "2601.06992",
"authors": [
"Yixi Zhou",
"Fan Zhang",
"Yu Chen",
"Haipeng Zhang",
"Preslav Nakov",
"Zhuohan Xie"
],
"categories": [
"cs.IR",
"cs.AI",
"cs.CL"
],
"license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
"title": "FinCARDS: Card-Based Analyst Reranking for Financial Document Question Answering",
"url": "https://arxiv.org/abs/2601.06992",
"version": "v1"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "177e5f45-c506-42a7-8054-d3d16c9b4283",
"id": "arXiv Dataset",
"type": "Model",
"variant": "snapshot-2026-01-17",
"version": "0.1.0"
},
"user_id": 1000002
}