Mahrus, Alif Muhammad (2026) Optimasi Portofolio Saham Berbasis Value-At-Risk Pada Indeks LQ45 Menggunakan Pendekatan Monte Carlo, Quantile Autoregressive, Dan Quantile Autoregressive Neural Network. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5003221035-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5003221035-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (14MB) | Request a copy |
Abstract
Pertumbuhan investor di pasar modal Indonesia menunjukkan meningkatnya aktivitas investasi, namun investor dihadapkan pada risiko akibat fluktuasi harga saham yang dinamis, sehingga pengukuran risiko menjadi aspek penting dalam pengambilan keputusan investasi. Penelitian ini bertujuan mengestimasi Value-at-Risk (VaR) pada return saham anggota indeks LQ45 menggunakan pendekatan Monte Carlo Simulation (MC), Quantile Autoregressive (QAR), dan Quantile Autoregressive Neural Network (QARNN) pada tingkat kuantil τ=1% dan τ=5%, mengevaluasi validitasnya melalui backtesting, serta membentuk portofolio optimal berbasis VaR. Hasil penelitian menunjukkan bahwa ketiga pendekatan mampu membedakan karakteristik risiko antar saham, di mana saham sektor energi dan pertambangan cenderung memiliki risiko lebih tinggi dibandingkan sektor perbankan dan konsumsi. Pada VaR-MC, performa terbaik diperoleh pada τ=5% dengan window 375 hari yang menghasilkan 22 saham valid berdasarkan backtesting. Pada VaR-QAR, seluruh 25 saham dinyatakan valid pada kedua tingkat kuantil sehingga menjadi metode dengan performa backtesting terbaik. Sementara itu, VaR-QARNN mampu menangkap pola nonlinear namun menghasilkan validitas yang lebih rendah, yaitu 10 saham valid pada τ=1% dan hanya 4 saham valid pada τ=5%. Optimasi portofolio dilakukan dalam dua skenario, yaitu menggunakan seluruh 25 emiten dan menggunakan emiten yang lolos backtesting. Hasil menunjukkan bahwa nilai Modified Sharpe Ratio (MSR) yang tinggi tidak selalu diikuti validitas estimasi risiko yang baik. Meskipun VaR-QARNN pada τ=5% menghasilkan MSR tertinggi pada skenario pertama, performa tersebut tidak dapat dipertahankan setelah mempertimbangkan hasil backtesting. Portofolio berbasis VaR-QAR pada τ=5% dipilih sebagai portofolio optimal terbaik karena memenuhi dua kriteria secara simultan, yaitu seluruh 25 emiten dinyatakan valid berdasarkan Uji Kupiec PoF dan menghasilkan MSR tertinggi di antara model dengan validitas penuh. Portofolio tersebut terdiri atas BBCA (20,01%), BBNI (20,00%), BBRI (19,98%), INDF (19,97%), UNVR (13,07%), dan PGAS (6,04%), dengan VaR portofolio sebesar 2,64% dan MSR sebesar 0,0666, yang mencerminkan keseimbangan terbaik antara validitas pengukuran risiko dan kinerja investasi.
=======================================================================================================================================
The growing number of investors in the Indonesian capital market reflects an increase in investment activity; however, investors are simultaneously exposed to risks arising from dynamic stock price fluctuations, making risk measurement a critical aspect of investment decision-making. This study aims to estimate Value-at-Risk (VaR) on the stock returns of LQ45 index constituents using three approaches, namely Monte Carlo Simulation (MC), Quantile Autoregressive (QAR), and Quantile Autoregressive Neural Network (QARNN), at quantile levels of τ=1% and τ=5%, to evaluate the validity of each estimation through backtesting, and to construct an optimal VaR-based portfolio. The results demonstrate that all three approaches are capable of distinguishing risk characteristics across stocks, whereby stocks in the energy and mining sectors tend to exhibit higher risk levels than those in the banking and consumer sectors. For VaR-MC, the best performance was achieved at τ=5% with a 375-day window, yielding 22 valid stocks based on backtesting. For VaR-QAR, all 25 stocks were declared valid at both quantile levels, establishing it as the method with the superior backtesting performance. In contrast, VaR-QARNN was able to capture nonlinear patterns in stock returns but produced comparatively lower validity, with only 10 valid stocks at τ=1% and 4 valid stocks at τ=5%. Portfolio optimization was conducted under two scenarios: one utilizing all 25 issuers and another restricted to issuers that passed the backtesting criterion. The findings indicate that a high Modified Sharpe Ratio (MSR) value does not necessarily correspond to sound risk estimation validity. Although VaR-QARNN at τ=5% yielded the highest MSR under the first scenario, this performance could not be sustained once backtesting results were taken into account. Consequently, the VaR-QAR-based portfolio at τ=5% was selected as the best optimal portfolio, as it simultaneously satisfied two criteria: all 25 issuers were declared valid under the Kupiec Proportion of Failures (PoF) Test, and it produced the highest MSR among all models with full validity. The resulting portfolio comprises BBCA (20.01%), BBNI (20.00%), BBRI (19.98%), INDF (19.97%), UNVR (13.07%), and PGAS (6.04%), with a portfolio VaR of 2.64% and an MSR of 0.0666, reflecting the most favorable balance between risk measurement validity and investment performance.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Monte Carlo, Optimasi Portofolio, QAR, QARNN, Value-at-Risk, Monte Carlo, Portfolio Optimization, QAR, QARNN, Value-at-Risk |
| Subjects: | H Social Sciences > HG Finance H Social Sciences > HG Finance > HG4529 Investment analysis H Social Sciences > HG Finance > HG4529.5 Portfolio management H Social Sciences > HG Finance > HG4910 Investments |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis |
| Depositing User: | Alif Muhammad Mahrus |
| Date Deposited: | 31 Jul 2026 02:06 |
| Last Modified: | 31 Jul 2026 02:06 |
| URI: | http://repository.its.ac.id/id/eprint/140479 |
Actions (login required)
 |
View Item |