Sparse Portfolio Optimization for US Pension Fund Quantitative Trading

Pension fund portfolio optimization is a critical task that involves managing risk and maximizing returns while adhering to operational constraints. To address the challenges of complexity and management costs, this paper proposes a novel sparse portfolio optimization framework. The key innovation lies in introducing an m-sparse constraint, which limits the number of active assets, significantly reducing management costs while maintaining performance. The framework combines multiple practical constraints, such as self-financing and long-only conditions, ensuring the feasibility and stability of the portfolio. To solve the non-convex optimization problem, we adapt the Proximal Gradient Algorithm (PGA), which guarantees global optimality with high computational efficiency. Experimental results show that the proposed method outperforms state-of-the-art algorithms in terms of Sharpe ratio and cumulative return, while also minimizing transaction costs. Our method provides a highly scalable and efficient solution for large-scale pension fund portfolio optimization, offering significant advantages in both performance and practicality.