CIFQA is a deterministic, tool‑grounded multi‑agent framework that separates language understanding from numerical execution for financial question answering. It assigns specialized agents for interpretation, routing, parameter extraction, computation planning, and response generation, while deterministic Python tools perform the calculations. On a fixed‑deposit benchmark, CIFQA achieves 95.54% accuracy on calculation‑intensive queries and 90.87% overall, outperforming larger LLM baselines and showing that architecture, not scale, drives numerical reliability.
By Kunjesh Parekh, Anil Kumar Tiwari, Divya Saxena
FinRAG-QA is a new benchmark dataset for financial question answering, featuring 999 practitioner-curated questions on 10 standardised indicators drawn from 209 annual and Pillar 3 reports of 24 major European and U.S. banks between 2019 and 2023. The dataset focuses on cross‑institutional retrieval over documents averaging 198k words, making it longer than any existing financial QA resource. Experiments on a multi‑stage Retrieval‑Augmented Generation pipeline show that contextual chunk enrichment and a retrieval‑optimised embedding model significantly improve NDCG@10, while a reasoning‑optimised generator boosts answer accuracy from 44.6% to 79.0% when the correct document is retrieved.
By Arianna Miola, Bruno Spaccavento, Lorenzo Silotto, Marco Bianchetti, Luca Cagliero
arXiv:2606. 17642v1 Announce Type: new Abstract: Financial multimodal reasoning requires agents to coordinate numerical computation, retrieval, visual interpretation, and temporal grounding across heterogeneous evidence sources.
By Pianran Guo, Pengcheng Zhou, Yucheng Jian, Shuhua Chen
The paper introduces Self-Improving Retrieval-Augmented Generation (RAG), a framework that splits document question answering into Retrieval, Reasoning, and Judge agents coordinated by an orchestrator. When the Judge scores an answer below a dynamic threshold, the system retries with broader retrieval, more careful prompting, and relaxed acceptance criteria, achieving 86% oracle-guided accuracy on FinanceBench with a 36.4% Lazarus Rate. The approach logs every decision with confidence scores, providing audit trails needed for regulated financial applications.
By Junjie Xiong, Shawheen Ghezavat, Aum Hirpara
arXiv:2608. 11047v1 Announce Type: new Abstract: While existing benchmarks have made substantial progress in evaluating LLMs across STEM domains, financial reasoning over structured data remains comparatively less explored.
By Alicia Larsen, Victoire Laurent, Aulia Kharis Rakhamsari, Lara Turgut, Nino Antulov-Fantulin
FinRCA-Bench is a synthetic benchmark designed to evaluate evidence retrieval and reasoning in financial AI systems, specifically for accounts‑payable‑to‑bank reconciliation. It contains 2,250 cases across 14 operational tables, with 1,500 injected failures in 15 causal categories and 750 hard‑negative cases, and hides root‑cause labels and evidence contracts to isolate retrieval performance. Experiments show that retrieval architecture dramatically affects accuracy, with structured retrieval methods like Typed Provenance Graph Retrieval vastly improving macro‑recall and exact‑class accuracy compared to dense semantic retrieval or classical ML.
arXiv:2603. 19225v3 Announce Type: replace-cross Abstract: Real-world financial decision-making is a challenging problem that requires reasoning over heterogeneous signals, including company fundamentals derived from regulatory filings and trading signals computed from price dynamics.
By Yogesh Agrawal, Aniruddha Dutta, Md Mahadi Hasan, Santu Karmaker, Aritra Dutta
FinRCA-Bench is a deterministic synthetic benchmark comprising 2,250 accounts‑payable‑to‑bank reconciliation cases that span 14 operational tables and include 1,500 injected failures across 15 causal categories. The benchmark hides root‑cause labels and record‑level evidence contracts from models, enabling independent evaluation of evidence retrieval versus reasoning accuracy. Experiments show that retrieval architecture dramatically influences performance, with retrieval improvements raising macro‑required‑record recall from 0.83% to 77.70% and exact 16‑class accuracy from 2.05% to 72.44%.
By Pratik Ghawate
arXiv:2609.25192v1 Announce Type: new
Abstract: Financial search is a highly demanding task for LLM agents, requiring not only a correct final answer but also temporally valid information retrieval,...
By Wenqing Wang, Haitao Xiang, Xinyi Zhao, Mingming Yin, Ying Zhong, Zhaoxin Huan, Qiheng Zhou, Jin Zhu, Xiaolu Zhang, Shi Chang, Jun Zhou
ASDA (Automated Skill Distillation and Adaptation) is a framework that improves large language models on financial reasoning tasks without fine‑tuning. It works by having a teacher model analyze a student’s failures, cluster errors, and generate structured skill artifacts—reasoning procedures, code templates, and worked examples—that are injected during inference. On the FAMMA benchmark, ASDA boosts arithmetic reasoning by up to 17.33% and non‑arithmetic reasoning by 5.95%, outperforming existing training‑free methods.
By Tik Yu Yim, Wenting Tan, Sum Yee Chan, Tak-Wah Lam, Siu Ming Yiu
The paper presents a retrieval‑augmented generation pipeline for answering regulatory compliance questions in finance. It builds a three‑stage retriever on LegalBERT and a compact 2B–12B generator served with 4‑bit quantization, achieving a Recall@10 of 0.774 on the ObliQA benchmark and improving answer quality via RAFT‑LoRA fine‑tuning. However, the adapted models fail to transfer to Australian case‑law questions, and a closed‑book model performs almost as well while lacking verifiable grounding.
By Tobias Deu{\ss}er, Abhishek Pillai, Aurelio F. Bariviera, Dhananjay Bhardwaj, Lorenz Sparrenberg, David Berghaus, Christian Bauckhage, Rafet Sifa
Finance Agent v2 (by Vals AI) has emerged as the reference benchmark for evaluating both Anthropic Claude and OpenAI ChatGPT frontier language models on financial tasks. However, it narrowly deals with periodic reporting from publicly traded companies (SEC 10-K and 10-Q filings), and its agentic harness relies on naive, unenriched chunk retrieval.