| Abstrak/Abstract |
The emergence of risk-based portfolio optimization was driven by the under-performance and constraints of meanvariance
(MV) analysis. The Risk Parity (RP) portfolio distributes capital to ensure each asset maintains an equal degree
of risk compared to the portfolio's total risk. The RP portfolio is a non-convex optimization problem that can be
addressed using conventional numerical methods, which may be inefficient and fail to yield a suitable solution. Using
stock price data of companies listed on the Indonesia Stock Exchange's IDX30 index, this study compares the Genetic
Algorithm (GA), a metaphor-based meta-heuristic algorithm based on the principles of biological evolution, with the
Successive Convex optimization for Risk Parity portfolio (SCRIP), which is based on a Successive Convex
Optimization (SCO) method, to determine the optimal solution for the long-only and the long-short RP portfolios. The
study demonstrates that GA produces a solution that slightly deviates from an equal-risk contribution solution. SCRIP
presents a solution that equally distributes risk. GA is highly efficient and could be enhanced to address the RP
portfolios that involve non-convex real-world constraints, which cannot be resolved with SCRIP. |
