Stock Recommendations for Individual Investors: A Temporal Graph Network Approach with Mean-Variance Efficient Sampling

The project is primarily an academic contribution (arXiv paper) rather than a production-grade tool. With 0 stars and only 5 forks over two years, it lacks any meaningful community adoption or developer ecosystem. The defensibility is low because the 'moat' consists solely of the specific algorithmic combination of Temporal Graph Networks and Mean-Variance sampling, which is easily reproducible by any quantitative researcher. Frontier labs (OpenAI, Anthropic) are unlikely to compete directly due to the regulatory and liability risks associated with providing financial advice, hence the 'low' frontier risk. However, platform domination risk is high because the natural home for this technology is within established brokerage platforms (Robinhood, Fidelity, E*Trade) or specialized fintech apps (Wealthfront, Betterment). These incumbents already possess the necessary user data and regulatory licenses to implement such features. The project serves as a solid reference for building a 'theoretically sound' recommender system, but lacks the data gravity or network effects required to survive as a standalone entity.