Alfiyani, Dewi (2026) Optimasi Portofolio Saham Menggunakan Non-Dominated Sorting Genetic Algorithm II Berbasis GBM-Monte Carlo Dengan Kendala Kardinalitas, Buy-In Threshold, Dan Roundlot. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5006221023-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5006221023-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (12MB) | Request a copy |
Abstract
Ketidakpastian pergerakan harga saham di Bursa Efek Indonesia menuntut pendekatan pemodelan yang mampu merepresentasikan sifat stokastik harga sekaligus mempertimbangkan kendala investasi yang realistis. Penelitian ini mengintegrasikan model Geometric Brownian Motion-Monte Carlo Simulation (GBM-MCS) untuk memprediksi harga saham dengan algoritma Non-Dominated Sorting Genetic Algorithm II (NSGA-II) untuk optimasi portofolio multiobjektif, menggunakan expected return dan Conditional Value at Risk (CVaR) sebagai fungsi objektif. Objek penelitian adalah survivor indeks LQ45 dengan data harga penutupan harian periode 1 Januari 2022 hingga 31 Desember 2025, yang dibagi menjadi 80% data in-sample dan 20% data out-sample. Optimasi portofolio memperhitungkan kendala kardinalitas, buy-in threshold, dan roundlot melalui sembilan kombinasi kendala, dan kinerja portofolio optimal dievaluasi menggunakan Sharpe Ratio, Sortino Ratio, dan Maximum Drawdown. Hasil penelitian menunjukkan bahwa model GBM-MCS menghasilkan akurasi prediksi yang lebih baik dibandingkan model GBM standar, ditandai dengan penurunan Mean Absolute Percentage Error (MAPE) pada 26 dari 27 saham, dengan penurunan tertinggi pada saham MEDC (17,87%), BBNI (12,35%), dan AMRT (12,03%). Dari sembilan kombinasi kendala yang diuji, tujuh kombinasi menghasilkan portofolio optimal yang feasible. Portofolio skenario 2 memberikan expected return, Sharpe Ratio, dan Sortino Ratio out-sample tertinggi (62,77%; 2,3735; 2,2905), namun menunjukkan lompatan kinerja yang besar terhadap hasil in-sample, sedangkan portofolio skenario 5 dan 7 menunjukkan konsistensi kinerja yang lebih baik dengan profil risiko lebih rendah, termasuk Maximum Drawdown terkecil sebesar -9,57% pada skenario 7. Penelitian ini menegaskan bahwa integrasi GBM-MCS dan NSGA-II dengan kendala praktis dapat mendukung pembentukan portofolio yang lebih efisien dan terukur bagi investor.
========================================================================================================================
The uncertainty of stock price movements on the Indonesia Stock Exchange calls for a modeling approach that captures the stochastic nature of prices while accounting for realistic investment constraints. This study integrates the Geometric Brownian Motion-Monte Carlo Simulation (GBM-MCS) model for stock price prediction with the Non-Dominated Sorting Genetic Algorithm II (NSGA-II) for multi-objective portfolio optimization, using expected return and Conditional Value at Risk (CVaR) as objective functions. The research object comprises 27 survivor stocks of the LQ45 index, using daily closing price data from January 1, 2022 to December 31, 2025, split into 80% in-sample and 20% out-sample data. Portfolio optimization incorporates cardinality, buy-in threshold, and roundlot constraints across nine constraint combinations, with optimal portfolio performance evaluated using the Sharpe Ratio, Sortino Ratio, and Maximum Drawdown. The results show that the GBM-MCS model produces better prediction accuracy than the standard GBM model, reducing the Mean Absolute Percentage Error (MAPE) in 26 of 27 stocks, with the largest reductions observed in MEDC (17,87%), BBNI (12,35%), and AMRT (12,03%). Of the nine constraint combinations tested, seven yielded feasible optimal portfolios. The portfolio from scenario 2 achieved the highest out-sample expected return, Sharpe Ratio, and Sortino Ratio (62,77%; 2,3735; 2,2905), but exhibited a large performance jump relative to its in-sample results, whereas portfolios from scenarios 5 and 7 showed more consistent performance with lower risk profiles, including the smallest Maximum Drawdown of -9,57% in scenario 7. This study confirms that integrating GBM-MCS and NSGA-II with practical constraints can support the construction of more efficient and measurable portfolios for investors.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Analisis investasi, Conditional Value at Risk, GBM-Monte Carlo Simulation, LQ45, manajemen risiko keuangan, NSGA-II, optimasi portofolio, pareto front, Conditional Value at Risk, Financial risk management, GBM-Monte Carlo Simulation, investment analysis, LQ45, NSGA-II, pareto front, portfolio optimization |
| Subjects: | H Social Sciences > HG Finance > HG4529.5 Portfolio management H Social Sciences > HG Finance > HG4915 Stocks--Prices Q Science > QA Mathematics > QA274.2 Stochastic analysis Q Science > QA Mathematics > QA402.5 Genetic algorithms. Interior-point methods. |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis |
| Depositing User: | Dewi Alfiyani |
| Date Deposited: | 17 Jul 2026 03:57 |
| Last Modified: | 17 Jul 2026 03:57 |
| URI: | http://repository.its.ac.id/id/eprint/135261 |
Actions (login required)
 |
View Item |