Optimasi Portofolio Saham dengan Genetic Algorithm Berbasis Graph Representation Learning

Nurhayati, Alfia (2026) Optimasi Portofolio Saham dengan Genetic Algorithm Berbasis Graph Representation Learning. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5002211084-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Investasi saham di pasar modal Indonesia, khususnya pada indeks LQ45, terus mengalami perkembangan yang dipengaruhi oleh dinamika pasar dan kemajuan teknologi komputasi. Seiring meningkatnya kompleksitas data keuangan, diperlukan metode optimasi yang mampu menangani hubungan non-linear antar aset. Penelitian ini mengusulkan sebuah pendekatan dalam pengelolaan portofolio dengan mengintegrasikan Graph Representation Learning (GRL) dan Genetic Algorithm (GA). Dalam kerangka kerja ini, GRL digunakan untuk mengekstraksi fitur dan memetakan struktur hubungan antar saham berdasarkan korelasi nilai return dalam bentuk graf, yang kemudian digunakan sebagai dasar diversifikasi. Selanjutnya, GA diterapkan untuk mengoptimalkan alokasi bobot portofolio melalui mekanisme seleksi, crossover, dan mutasi guna mencapai keseimbangan risiko dan return.
Hasil analisis menunjukkan bahwa integrasi GRL+GA pada saham-saham indeks LQ45 menghasilkan kinerja yang kompetitif dengan nilai Sharpe Ratio sebesar 1.4069. Angka ini lebih tinggi dibandingkan penggunaan Genetic Algorithm standar tanpa informasi graf yang menghasilkan Sharpe Ratio sebesar 1.3920. Meskipun metode dengan Mean-Variance Optimization (MVO) secara matematis menghasilkan Sharpe Ratio lebih tinggi sebesar 1.4102, pendekatan GRL+GA menawarkan keunggulan dalam menghasilkan alokasi bobot yang lebih stabil dan terdiversifikasi secara sistematis tanpa konsentrasi aset yang ekstrem. Temuan ini mengindikasikan bahwa pemanfaatan hubungan struktural antar aset berbasis data graf dapat menjadi alternatif bagi investor dalam menghadapi kondisi pasar yang dinamis di Indonesia.
==================================================================================================================================
Investment in stocks on the Indonesian capital market, particularly in the LQ45 index, continues to experience growth influenced by market dynamics and advances in computing technology. As financial data becomes increasingly complex, optimization methods capable of handling non-linear relationships between assets are needed. This study proposes an approach to portfolio management by integrating Graph Representation Learning (GRL) and Genetic Algorithm (GA). In this framework, GRL is used to extract features and map the relationship structure between stocks based on the correlation of return values in the form of graphs, which are then used as the basis for diversification. Furthermore, GA is applied to optimize portfolio weight allocation through selection, crossover, and mutation mechanisms to achieve a balance between risk and return. The analysis results show that the integration of GRL+GA in LQ45 index stocks produces competitive performance with a Sharpe Ratio value of 1.4069. This figure is higher than the use of a standard Genetic Algorithm without graph information, which produces a Sharpe Ratio of 1.3920. Although the method with Mean-Variance Optimization (MVO) mathematically produces higher Sharpe Ratio of 1.4102, the GRL+GA approach offers the advantage of producing a more stable and systematically diversified weight allocation without extreme asset concentration. These findings indicate that utilizing data-based structural relationships between assets can be an alternative for investors in facing dynamic market conditions in Indonesia.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Genetic Algorithm, Graph Representation Learning, Indeks LQ45, Optimasi Portofolio, Sharpe Ratio, Genetic Algorithm, Graph Representation Learning, LQ45 Index, Portfolio Optimization, Sharpe Ratio
Subjects:	Q Science > QA Mathematics > QA402.5 Genetic algorithms. Interior-point methods.
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User:	Alfia Nurhayati
Date Deposited:	02 Feb 2026 05:08
Last Modified:	02 Feb 2026 05:08
URI:	http://repository.its.ac.id/id/eprint/131742

Actions (login required)

View Item