Zero-Shot Time Series Foundation Models for Annual Institutional Forecasting Under Data Sparsity

This project is a research artifact (likely a code release for the cited Arxiv paper) focused on a highly specific domain: annual institutional enrollment forecasting. While it addresses a real-world pain point—data sparsity and regime shifts in institutional planning—it currently lacks any meaningful adoption (0 stars). Its primary value lies in the benchmark of existing Foundation Models (like Amazon's Chronos or Google's TimesFM) against classical baselines. The 'leakage-safe covariate' approach is a useful technical contribution but does not constitute a moat. Defensibility is low because the core logic relies on third-party models and standard time-series preprocessing. Frontier labs (Google, Amazon) are already dominating the underlying model space, and specialized forecasting firms (e.g., Nixtla) or ERP providers (e.g., Workday, Ellucian) are the natural homes for this functionality. The project is more of a validation of TSFM utility in niche sectors than a standalone defensible project.