RA research statement draft

I am interested in reliable prediction and decision support under limited information. My current work starts from financial volatility forecasting, where data and high-quality information are expensive. The SRM project explores whether geometric analog retrieval can provide a simple, auditable, and useful forecasting model without relying on an unnecessarily large information set.

I am also beginning to study world models for financial time series. The public prototype is still early and its model quality is not yet a finished result, but the project has helped me think about latent dynamics, probabilistic future paths, shared structure across assets, and forward-only evaluation.

Together with Associate Professor Yirong Huang, I am interested in whether AI agents can support financial decisions without hiding their information boundary or uncertainty. In the longer term, I hope to move beyond forecasting toward the decision problems that follow from forecasts.

I am applying for an RA position to strengthen my ability to identify and formulate research ideas, receive systematic research training, and learn more about the path toward a PhD or MPhil. I hope to work with others on useful, well-defined problems and contribute to high-quality papers. The lab’s focus on predictive thinking, world models, and general agents is a natural environment for this transition.

I also have a parallel interest in Schubert calculus and algebraic geometry. If the opportunity arises, I would like to explore whether its structured combinatorial representations can inspire general algorithms for learning or reasoning.

Author checks before submission

• Confirm the exact RA dates and availability.
• Add one or two concrete target problems from the lab.
• Confirm manuscript and arXiv wording for the SRM project.
• Confirm whether to mention Schubert-calculus/ML connections in the final one-page statement.