Researchers have developed PortBench, a new benchmark designed to evaluate Large Language Models (LLMs) in portfolio management. Existing benchmarks fail to account for crucial cross-asset correlations and the full decision-making pipeline. PortBench addresses these gaps with a static question-answering dataset and a dynamic five-stage allocation pipeline, introducing new metrics to assess portfolio diversification and error compounding. Evaluations of ten frontier LLMs revealed that 90% performed worse than a simple equal-weight strategy, with even compliant models experiencing significant drawdowns during stress periods. AI
IMPACT Highlights critical limitations of current LLMs in complex financial decision-making, necessitating further research for real-world application.
RANK_REASON This is a research paper introducing a new benchmark for evaluating LLMs in a specific domain. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
