CD Tesis
Model Mean-Expected Shortfall Untuk Optimisasi Portofolio Saham Dengan Support Vector Regression Serta Aplikasinya Pada Pasar Modal
ABSTRACT
RISTIFANI ULFATMI NIM. 2210246884, Mean-Expected Shortfall Model
Model for Stock Portfolio Optimization with Support Vector Regression
and its Application to the Capital Market, supervised by M. D. H. Gamal
and Arisman Adnan.
Public awareness of the importance of investment is increasing in line with
advances in technology and information. In the context of stock investment,
portfolio diversification is one effective strategy for reducing risk. This study
discusses stock portfolio optimization using the Mean-Expected Shortfall (Mean-
ES) approach due to its ability to capture extreme risk above the Value at Risk
(VaR) threshold. To support the accurate estimation of portfolio returns, the
Support Vector Regression (SVR) model is used because it has good predictive
performance on non-linear and complex data. Five plantation sector stocks AALI,
SIMP, SGRO, DSNG, and SMAR yield optimal weights of 12%, 27%, 38%, 2%,
and 21%, respectively. SVR with linear and RBF kernels were used for portfolio
return forecasting with optimal stock weights, as both kernels produced excellent
regression models with low Mean Absolute Percentage Error parameters and high
determination coefficients for both kernels. The portfolio return forecasts using
SVR differ from actual results in the capital market. This may occur due to
market fluctuations, economic policies, and other external factors that cannot be
predicted and influence stock prices. Nevertheless, the Mean-ES model is capable
of providing highly accurate risk estimates even under extreme market conditions.
Keywords: Mean-ES, SVR, portfolio optimization, capital market
Tidak tersedia versi lain