Songyang Li

Title: Songyang Li and Machine Learning for Portfolio Optimization.

Introduction

Songyang Li is an inventor based in Somerville, Massachusetts, and is associated with FMR Corp. His patent work applies machine learning to financial portfolio simulation and optimization.

Latest Patents

Li is named on 1 patent: “Machine Learning Portfolio Simulating and Optimizing Apparatuses, Methods and Systems.” The invention describes a system that processes requests for machine-learning simulations, decision-tree ensemble training, expected-return calculations, portfolio construction, scenario creation, and visualization of portfolio returns.

The system uses neural networks to encode historical market risk factors into latent variables and decode them into simulated market scenarios. Its transfer layer allows the latent variables to represent different distributions and dependency structures, helping the simulated scenarios reflect the dynamic dependencies and volatility found in historical market data.

Career Highlights

Li’s disclosed patent work focuses on applying machine learning techniques to portfolio analysis. The system brings together market scenario simulation, expected-return estimation, and portfolio construction in a unified framework.

Collaborations

The provided information lists Aaron Gao and Samarjit Walia as Li’s coworkers. Their specific roles in the patented invention are not stated.

Conclusion

Songyang Li’s patent describes a machine-learning framework for generating market scenarios and supporting portfolio analysis, simulation, and optimization.