QuantEval: A Benchmark for Financial Quantitative Tasks in Large Language Models
QuantEval is a benchmark testing Large Language Models (LLMs) across three quantitative finance skills: knowledge QA, mathematical reasoning, and strategy coding. It uses a realistic CTA-style backtesting framework to evaluate if LLMs can complete the full compute-implement-validate loop required in quantitative finance. Results show significant gaps between LLM performance and human experts, especially in complex reasoning and coding tasks.