Tongren Xiao (肖童仁)

Tongren Xiao (肖童仁)

Mathematics undergraduate working on reliable AI, structured decision support, and uncertainty-aware financial modeling.

Curriculum Vitae

Download the current English CV (PDF)

Research interests

AI agents; structured decision support; world models; financial AI; time-series forecasting.

Education

Sun Yat-sen University, School of Mathematics, Guangzhou, China

Undergraduate Student in Mathematics, 2023-September 2027 (expected)

Research mentorship

Quantitative Finance, 2024-2026 Advisors: Associate Professor Yirong Huang and Associate Professor Yujun Lian, Sun Yat-sen University

Schubert Calculus and Algebraic Geometry, 2024-2026 Advisor: Professor Changzheng Li, Sun Yat-sen University

Selected research

Shape Is Useful: Geometric Analog Forecasting of Realized Volatility. With Shuhang Chen; advised by Associate Professor Yujun Lian. Manuscript, 2026; arXiv preprint forthcoming. Developed the Shape Retrieval Model, a nonparametric forecaster that retrieves historical volatility paths through five geometric-similarity channels with Transformer-learned weights; strong performance at short and medium horizons.

SRM Forecasting Agent and Web Research Interface. 2026. Designed an auditable agent around the Shape Retrieval Model, with validated tool calls, information-boundary checks, neighbor explanations, and reproducible JSON traces. In discussion with the Headline Arena team about integrating its public forward-evaluation API.

Reducing Underestimation Risk in Volatility Forecasts: A Direction-Aware Post-Processing Method. With Yihua Yue; advised by Associate Professor Yirong Huang. Manuscript and research presentation, 2026; arXiv preprint forthcoming. Developed a residual-tail calibration layer for HAR/GHAR and IV-augmented forecasts, using leakage-controlled rolling protocols to reduce underestimation frequency and magnitude without redesigning the base model.

A Comparative Study of Hybrid LSTM Frameworks for Volatility Forecasting in the NASDAQ-100 and S&P 500 Markets. Yihua Yue, Rong Jia, Zimeng Lv, Tongren Xiao, and Ke Li. Manuscript, 2025. Compared hybrid LSTM-GARCH frameworks using realized volatility, GARCH-family forecasts, and implied-volatility information.

A Passive Recurrent State-Space World Model for Multi-Asset Realized Volatility. With Shuhang Chen. Preliminary research prototype and early-stage manuscript, 2026; the model is not yet mature. Built an anchor-conditioned RSSM with a causal HAR-IV level component and market-shared and asset-specific latent states; the audited pilot establishes feasibility, not benchmark superiority.

Graham Positivity of Quantum Double Schubert Polynomials. With Yihua Yue; advised by Professor Changzheng Li. Manuscript, 2026. Developed a geometric proof of coefficient positivity in the product expansion of quantum double Schubert polynomials; the working draft is under discussion with Matthew J. Samuel.

Submatrix Selection for Matrices with Orthonormal Columns. Research manuscript, 2026, on the nearly three-decade-old Goreinov-Tyrtyshnikov-Zamarashkin conjecture. Obtained improved lower bounds for $k=3$ and proved structured cases; the unrestricted case remains open. Currently discussing and checking proof details with Richik Sengupta, coauthor of the latest $k=2$ result.

Coursework and academic teaching

Selected advanced coursework: Selected Readings of Academic Articles (graduate-level, 98/100), Harmonic Analysis on Groups (graduate-level, 92.5/100), Algebraic Topology (honors, 83/100), and Stochastic Processes (interdisciplinary). In Selected Readings of Academic Articles, delivered the main lectures on Chapters 1 and 3 of Emily Riehl’s Category Theory in Context.

International academic training

Introduction to Option Pricing, 7 March-25 May 2025. International research-based learning course taught by Johannes Ruf; grade: 81.50.

Honors and projects