xiry.

Learning · Article or course

Bridging Language Models and Financial Analysis

Survey paper exploring large language model applications and limitations in finance

From the Xiry library · Updated 10 October 2026

Visit arxiv.org ↗

A 28-page survey paper on arXiv (arXiv:2503.22693) covering the integration of large language models into quantitative finance and trading. It examines how modern LLM techniques process multifaceted financial inputs like text, tables, and charts, while detailing major challenges including hallucinations and math errors. It serves as a thorough primer for researchers and quantitative practitioners wanting to apply AI to financial analysis.

What it's for

  • Understand the state of LLM adoption and research in quantitative finance
  • Identify common failure modes like mathematical errors and hallucinations in financial prompts
  • Design prompts using in-context learning to improve financial sentiment analysis

Features

  • Comprehensive review of recent LLM architectures applied to finance
  • Analysis of multimodal processing across text, tables, and financial charts
  • In-depth breakdown of failure cases including calculation errors and factual hallucinations
  • Survey of in-context learning techniques for financial sentiment scoring
  • Discussion of research gaps between NLP developments and institutional finance adoption

Pricing

Free open-access paper on arXiv

Details

Type
Article or course
Category
Learning
Topic
Research Papers
Open source
Yes
Platforms
Web
For
Quantitative analysts, financial data scientists, and AI researchers working in finance.
Tags
arxiv, finance, quantitative, survey, llms, research

Why it was saved

  • @quantscience_, 5 April 2025: Useful starting point for applying LLMs to financial data, and for knowing their weaknesses (hallucination, math reasoning)

Related

Updated