Portfolio Optimization of Indonesian Banking Stocks Using Robust Optimization

Battle Count: 175

The paper addresses portfolio construction and asset allocation, which are foundational to quantitative trading strategies. However, it focuses on static portfolio optimization rather than dynamic trading signals, execution algorithms, or high-frequency strategies. The robust optimization framework is relevant for risk-aware portfolio construction in quantitative trading systems, particularly for emerging market applications. The findings about moving-window vs. bootstrapping uncertainty sets could inform parameter estimation choices in trading systems.

The methodology involves standard quadratic programming (solved via CPLEX), moving-window computation (straightforward sliding window), and block bootstrapping (moderate complexity with block resampling). The mathematical formulation is well-defined and the algorithms are clearly presented. Main complexity lies in correctly implementing the uncertainty set construction and ensuring the robust reformulation is properly linearized. No deep learning or complex iterative algorithms are required.

The paper provides clear algorithmic descriptions (Algorithm 1 and 2), specifies software (MATLAB R2021b, CPLEX 12.10.0), lists all 45 stock codes, and defines parameter choices (K=90, N_boot=1000, α=0.05, γ={5,50,100}). However, no code repository is provided, and the exact data preprocessing steps and CPLEX solver settings are not fully detailed. Data is sourced from Yahoo Finance (publicly available).