BigFinanceBench: A Workflow-Grounded Benchmark for Financial-Research Agents
2026-06-02 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors created BigFinanceBench, a new test for financial research that checks not just the final answer but all the steps taken to reach it. This helps ensure that others can verify exactly how a financial conclusion was made, including sources and assumptions. They found that current AI systems still struggle to fully match expert financial reasoning, scoring less than 60% on detailed steps. Their benchmark helps identify which parts of the process need improvement, beyond just looking at whether the final answer is right or wrong.
financial researchbenchmarkrubricpartial-credit evaluationworkflowauditabilityfinancial analysisAI evaluationderivationmodel capability
Authors
Alex Wang, Georg Meinhardt, Jacob Katz, Joseph H. Kim, Pratyush K. Chaudhary, Chase Blagden, Eric Xu
Abstract
Financial-research answers are decision-relevant only when another analyst can audit how they were produced: which source was chosen, which period and accounting definition were used, which assumptions were made, and how the calculation was performed. Existing finance benchmarks largely evaluate isolated subskills or final answers, leaving the auditable derivation itself under-measured. We introduce BigFinanceBench, a 928-item expert-authored benchmark of open-ended financial-research tasks in which each item pairs a ground-truth reference answer with a point-weighted rubric that decomposes the derivation into independently checkable steps. BigFinanceBench is workflow-grounded in that it evaluates the full derivation rather than only the final output. Across 36,241 rubric points, the benchmark supports partial-credit evaluation and localization of failures across the analyst workflow. Evaluating ten current frontier and open-weight agents, we find substantial headroom: the best system reaches only 58.8% rubric score, final-answer accuracy is a useful but lossy proxy for derivation quality, and model capability varies non-uniformly across financial workflows.