Benchmarks Are Not Validation: A System-Level View of Financial LLM Applications

arXiv:2607.28840 · cs.CL, cs.SE · Submitted 2026-07-30 · Read on arXiv

Burak Payzun, İrem Demirtaş, Simona Scala, Elena Ferretti, Seçil Arslan

cs.CL, cs.SE

Submitted: 2026-07-30

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Benchmark — 44× in this paper · explained in 2 episodes
• Trace — 15× in this paper · explained in 2 episodes
• RAG — 14× in this paper · explained in 3 episodes
• LLM-as-a-Judge — 10× in this paper · explained in 4 episodes
• LLM judge — 7× in this paper · explained in 2 episodes
• Accuracy — 6× in this paper · explained in 1 episode
• Foundation Model — 6× in this paper · explained in 3 episodes
• Failure Modes — 5× in this paper · explained in 3 episodes
• Faithfulness — 5× in this paper · explained in 2 episodes
• Grounding — 5× in this paper · explained in 3 episodes
• Robustness — 5× in this paper · explained in 6 episodes
• Benchmarking — 4× in this paper · explained in 4 episodes
• Hallucination — 4× in this paper · explained in 4 episodes
• Ablation — 3× in this paper · explained in 1 episode
• Agentic System — 2× in this paper · explained in 2 episodes
• Calibration — 2× in this paper · explained in 9 episodes
• Consistency — 2× in this paper · explained in 3 episodes
• Explainability — 2× in this paper · explained in 4 episodes
• Fine-Tuning — 2× in this paper · explained in 5 episodes
• Foundation Models — 2× in this paper · explained in 8 episodes
• Human-in-the-loop — 2× in this paper · explained in 4 episodes
• Adversarial Robustness — 1× in this paper · explained in 4 episodes
• Augmentation — 1× in this paper · explained in 2 episodes
• Core — 1× in this paper · explained in 2 episodes
• Data Augmentation — 1× in this paper · explained in 10 episodes
• Distribution Shift — 1× in this paper · explained in 12 episodes
• Domain Adaptation — 1× in this paper · explained in 5 episodes
• Editor — 1× in this paper · explained in 2 episodes
• Evaluation Protocol — 1× in this paper · explained in 2 episodes
• Feedback Mechanisms — 1× in this paper · explained in 2 episodes
• Generalization — 1× in this paper · explained in 14 episodes
• Generator — 1× in this paper · explained in 2 episodes
• Judge Reliability — 1× in this paper · explained in 2 episodes
• Model Dependence — 1× in this paper · explained in 3 episodes
• Multi-Agent System — 1× in this paper · explained in 7 episodes
• Non-stationarity — 1× in this paper · explained in 2 episodes
• Overfitting — 1× in this paper · explained in 2 episodes
• Position Bias — 1× in this paper · explained in 2 episodes
• Prompt Engineering — 1× in this paper · explained in 6 episodes
• Task Decomposition — 1× in this paper · explained in 2 episodes
• Verification — 1× in this paper · explained in 2 episodes

Sources

• Advancing Financial Engineering with Foundation Models: Progress, Applications, and Challenges
• FinanceBench: A New Benchmark for Financial Question Answering
• GAIA: a benchmark for General AI Assistants
• Replacing Judges with Juries: Evaluating LLM Generations with a Panel of Diverse Models

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering