Why analyst narratives still matter in an AI world

Cristina Joos, Research Analyst

Post feature

The way institutional investors consume sell-side research is evolving. Analyst reports are being ingested by AI systems, and the numerical outputs are being priced into markets faster than ever. With this in mind, does the written content of analyst reports still carry investment value? On this topic, we highlight a paper that asks exactly this and discuss relevant datasets across broker research and estimates.

LITERATURE

We examine the 2025 paper ‘Do Sell-side Analyst Reports Have Investment Value?’ by Linying Lv, published as a working paper to SSRN in April 2025 and last revised in August 2025.

QUICK VIEW
  • The paper applies LLMs to convert analyst report narratives into numerical representations, then uses machine learning models to predict 12-month stock returns and construct tradable portfolios.
  • The paper finds that analyst narratives contain investment-relevant information that is not fully captured in traditional quantitative measures, generating a measurable alpha.
  • The findings point to a critical gap in how investment teams have historically assessed and extracted value from broker research.