Guiding Real-Time LLM Market Analytics for the UK Financial Industry
Project Summary
Automation of reporting, research and signal generation carry significant risks and opportunities. Reports generated by LLMs need to be actionable, explainable, inter-operable and compliant to deliver benefits for firms, investors, consumers and the economy. AI’s speed and scale need harnessing while safeguarding credibility and trust—supporting transparency, accessibility and inclusivity.
A rigorous approach to prompt engineering and data governance is business-critical for the UK financial sector given vulnerability to hallucinations and injections—producing plausible but incorrect information, and sabotage.
In collaboration with TT International, a specialist investment manager with over 35 years’ experience, this pilot will take a quasi-scientific approach to blended (price and news based) reporting protocols across multiple assets.
This will involve sandbox testing with live and historical data, co-mentoring with TT’s AI lead, and research into AI ‘co-analysts’, and development of a prompt playbook, contributing to UK financial industry best-practice guidelines for LLMs. The outcome is a research paper with policy recommendations for LLM prompting, templates and associated library development—adoptable and scalable industry-wide —alongside experimental evidence on how explanation quality and bias mitigation could influence, e.g. investor trust. These insights will inform best-practice standards and regulatory guidance for transparent, compliant AI use in financial markets.
Meet The Team
Samuel Alexander Beatson
Nottingham University Business School
Erdinc Akyildirim
Nottingham University Business School
Chandra Sekhar Mangipudi
Nottingham University Business School