UCL MSc Project
Research Proposals
Expected project timeline: summer 2027
Human oversight and agent autonomy in EM allocation research
Which research decisions can be delegated to an agent while maintaining quality and reducing analyst effort?
The project examines how analysts and LLM-based agents share decisions in emerging markets equity allocation. Its aim is to identify where delegation is useful, where human approval is needed, and the trade-offs in quality, time and cost.
Approach
Compare four autonomy levels across five research stages. Test selected workflows on held-out cases, then freeze a policy for prospective testing with live information and paper portfolios.
Expected output
A quantitative delegation map and tested escalation rules, with uncertainty and unresolved boundaries made explicit. Investment returns are a secondary measure.
Further proposals will be added here.
I welcome practitioner feedback on research workflows, quality standards and where human judgement matters.