Agentic Reasoning has become a transformative force in financial analysis due to its ability to integrate large-scale information and generate reliable and accurate content. However, when handling complex real-world problems, different agents still show significant performance variation. In this work, we design Finance-LaTeX SKILL, a skill for synthesizing financial documents with complex layouts based on expert knowledge. Using an agent workflow built on this skill, we generate 2,000 professional financial documents along with 6,000 high-quality question-answer pairs. To evaluate the overall capability of agents, we introduce FinanceComplexQA, a comprehensive open-ended generation benchmark for financial documents that closely resembles real-world scenarios. It contains 2,026 deep research tasks targeting 1009 financial documents. FinanceComplexQA has 8 key features: bilingual support; coverage of six mainstream scenarios and seven tasks; expert-level document reasoning questions; deep research of complex layouts; relatively stable and permanent reference answers; and precise evaluation through an Agent-as-a-Judge with multiple evaluation metrics. Using FinanceComplexQA, we conduct a comprehensive evaluation of leading RAG systems and agentic reasoning tools for financial document QA. Through identifying and analyzing failure cases, we provide an in-depth study of their capabilities in numerical computation, multi-hop reasoning, content summarization, and industry analysis.</p>\n","updatedAt":"2026-07-24T11:30:27.623Z","author":{"_id":"66c596fccf0439733e38913f","avatarUrl":"/avatars/f4805286596305996613689868c57fa2.svg","fullname":"Xianfu Cheng","name":"BuaaCXF","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":7,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8848273754119873},"editors":["BuaaCXF"],"editorAvatarUrls":["/avatars/f4805286596305996613689868c57fa2.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2607.19238","authors":[{"_id":"6a63434f501b0a811675372f","name":"Xianfu Cheng","hidden":false},{"_id":"6a63434f501b0a8116753730","name":"Shiwei Zhang","hidden":false},{"_id":"6a63434f501b0a8116753731","name":"Jiyu Zhao","hidden":false},{"_id":"6a63434f501b0a8116753732","name":"Jian Yang","hidden":false},{"_id":"6a63434f501b0a8116753733","name":"Xinyuan Wang","hidden":false},{"_id":"6a63434f501b0a8116753734","name":"Ming Zhou","hidden":false},{"_id":"6a63434f501b0a8116753735","name":"Weixiao Zhou","hidden":false},{"_id":"6a63434f501b0a8116753736","name":"Xiangyuan Guan","hidden":false},{"_id":"6a63434f501b0a8116753737","name":"Xiang Li","hidden":false},{"_id":"6a63434f501b0a8116753738","name":"Zhenhe Wu","hidden":false},{"_id":"6a63434f501b0a8116753739","name":"Ziyi Ni","hidden":false},{"_id":"6a63434f501b0a811675373a","name":"Zhoujun Li","hidden":false},{"_id":"6a63434f501b0a811675373b","name":"Bingjing Xu","hidden":false}],"publishedAt":"2026-07-21T00:00:00.000Z","submittedOnDailyAt":"2026-07-24T00:00:00.000Z","title":"FinanceComplexQA: Benchmarking Agentic Reasoning on Industrial-grade Financial Documents","submittedOnDailyBy":{"_id":"66c596fccf0439733e38913f","avatarUrl":"/avatars/f4805286596305996613689868c57fa2.svg","isPro":false,"fullname":"Xianfu Cheng","user":"BuaaCXF","type":"user","name":"BuaaCXF"},"summary":"Agentic Reasoning has become a transformative force in financial analysis due to its ability to integrate large-scale information and generate reliable and accurate content. However, when handling complex real-world problems, different agents still show significant performance variation. In this work, we design Finance-LaTeX SKILL, a skill for synthesizing financial documents with complex layouts based on expert knowledge. Using an agent workflow built on this skill, we generate 2,000 professional financial documents along with 6,000 high-quality question-answer pairs. To evaluate the overall capability of agents, we introduce FinanceComplexQA, a comprehensive open-ended generation benchmark for financial documents that closely resembles real-world scenarios. It contains 2,026 deep research tasks targeting 1009 financial documents. FinanceComplexQA has 8 key features: bilingual support; coverage of six mainstream scenarios and seven tasks; expert-level document reasoning questions; deep research of complex layouts; relatively stable and permanent reference answers; and precise evaluation through an Agent-as-a-Judge with multiple evaluation metrics. Using FinanceComplexQA, we conduct a comprehensive evaluation of leading RAG systems and agentic reasoning tools for financial document QA. Through identifying and analyzing failure cases, we provide an in-depth study of their capabilities in numerical computation, multi-hop reasoning, content summarization, and industry analysis.","upvotes":4,"discussionId":"6a63434f501b0a811675373c","projectPage":"https://huggingface.co/datasets/Multilingual-Multimodal-NLP/FinanceComplexQA","githubRepo":"https://github.com/buaacxf/FinanceComplexQA","githubRepoAddedBy":"user","githubStars":1,"organization":{"_id":"63ba7720fc454697637969f1","name":"Beihang","fullname":"Beihang University","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/63ba7666c138c8f2b7844b58/n98lZU9VWxYgWIkzE_6o4.jpeg"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"66c596fccf0439733e38913f","avatarUrl":"/avatars/f4805286596305996613689868c57fa2.svg","isPro":false,"fullname":"Xianfu Cheng","user":"BuaaCXF","type":"user"},{"_id":"69bf5ef49ac7b84490b5096c","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/A0b-T9AqmIPHHRaSrnB1r.png","isPro":false,"fullname":"Ruoxi Tang","user":"james-gonzalezh","type":"user"},{"_id":"63f7767fbd28622c9b9915e9","avatarUrl":"/avatars/d2c624d2815cb0effb0c26a7687b4388.svg","isPro":false,"fullname":"Ziyi Ni","user":"Nicole-Yi","type":"user"},{"_id":"63ac5701c21e60a3e9b58aa7","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63ac5701c21e60a3e9b58aa7/g6EX7diOpuA94R2ab-rZC.png","isPro":true,"fullname":"Dipankar Sarkar","user":"dipankarsarkar","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"63ba7720fc454697637969f1","name":"Beihang","fullname":"Beihang University","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/63ba7666c138c8f2b7844b58/n98lZU9VWxYgWIkzE_6o4.jpeg"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2607/2607.19238.md","query":{}}">
FinanceComplexQA: Benchmarking Agentic Reasoning on Industrial-grade Financial Documents
Abstract
Agentic Reasoning has become a transformative force in financial analysis due to its ability to integrate large-scale information and generate reliable and accurate content. However, when handling complex real-world problems, different agents still show significant performance variation. In this work, we design Finance-LaTeX SKILL, a skill for synthesizing financial documents with complex layouts based on expert knowledge. Using an agent workflow built on this skill, we generate 2,000 professional financial documents along with 6,000 high-quality question-answer pairs. To evaluate the overall capability of agents, we introduce FinanceComplexQA, a comprehensive open-ended generation benchmark for financial documents that closely resembles real-world scenarios. It contains 2,026 deep research tasks targeting 1009 financial documents. FinanceComplexQA has 8 key features: bilingual support; coverage of six mainstream scenarios and seven tasks; expert-level document reasoning questions; deep research of complex layouts; relatively stable and permanent reference answers; and precise evaluation through an Agent-as-a-Judge with multiple evaluation metrics. Using FinanceComplexQA, we conduct a comprehensive evaluation of leading RAG systems and agentic reasoning tools for financial document QA. Through identifying and analyzing failure cases, we provide an in-depth study of their capabilities in numerical computation, multi-hop reasoning, content summarization, and industry analysis.
Community
Agentic Reasoning has become a transformative force in financial analysis due to its ability to integrate large-scale information and generate reliable and accurate content. However, when handling complex real-world problems, different agents still show significant performance variation. In this work, we design Finance-LaTeX SKILL, a skill for synthesizing financial documents with complex layouts based on expert knowledge. Using an agent workflow built on this skill, we generate 2,000 professional financial documents along with 6,000 high-quality question-answer pairs. To evaluate the overall capability of agents, we introduce FinanceComplexQA, a comprehensive open-ended generation benchmark for financial documents that closely resembles real-world scenarios. It contains 2,026 deep research tasks targeting 1009 financial documents. FinanceComplexQA has 8 key features: bilingual support; coverage of six mainstream scenarios and seven tasks; expert-level document reasoning questions; deep research of complex layouts; relatively stable and permanent reference answers; and precise evaluation through an Agent-as-a-Judge with multiple evaluation metrics. Using FinanceComplexQA, we conduct a comprehensive evaluation of leading RAG systems and agentic reasoning tools for financial document QA. Through identifying and analyzing failure cases, we provide an in-depth study of their capabilities in numerical computation, multi-hop reasoning, content summarization, and industry analysis.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2607.19238 in a model README.md to link it from this page.
Cite arxiv.org/abs/2607.19238 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.