Estimasi Risiko Portofolio Optimal Single Index Model Berdasarkan Peramalan Hybrid Autoregressive Integrated Moving Average-Long Short Term Memory

Setyadi, Angelica Benedict (2025) Estimasi Risiko Portofolio Optimal Single Index Model Berdasarkan Peramalan Hybrid Autoregressive Integrated Moving Average-Long Short Term Memory. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5006211021-Undergraduate_Theses.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Pasar saham merupakan salah satu instrumen investasi yang menjanjikan karena potensi keuntungannya yang tinggi, meskipun dibarengi dengan risiko yang signifikan. Oleh karena itu, peramalan harga saham dan pembentukan portofolio yang optimal menjadi aspek krusial dalam mendukung pengambilan keputusan investasi yang lebih akurat dan efisien. Penelitian ini bertujuan untuk (1) meramalkan harga saham IDX High Dividend 20 (IDXHIDIV20) menggunakan metode hybrid ARIMA-LSTM, (2) membentuk portofolio optimal dengan pendekatan Single Index Model (SIM), serta (3) mengestimasi risiko portofolio menggunakan Value at Risk (VaR) dan Conditional Value at Risk (CVaR) berbasis simulasi Monte Carlo. Data yang digunakan adalah saham-saham ANTM, ASII, BBCA, BMRI, dan TLKM yang tergabung secara konsisten dalam indeks IDXHIDIV20 pada periode 3 Januari 2022 hingga 3 Januari 2025. Metode hybrid ARIMA-LSTM diterapkan karena mampu menggabungkan kekuatan ARIMA dalam menangkap pola linier dan LSTM dalam memodelkan pola non-linier pada data deret waktu. Hasil prediksi menunjukkan bahwa model hybrid menghasilkan nilai MAPE yang sangat baik, berkisar antara 0,927293% hingga 3,702155%, dengan akurasi yang dipengaruhi oleh jumlah neuron pada arsitektur LSTM. Selanjutnya, portofolio optimal SIM terdiri dari dua saham, yaitu BBCA (32,43%) dan BMRI (67,57%). Portofolio ini memiliki expected return yang kompetitif, meskipun masih terdapat satu saham, yaitu BMRI, yang memiliki expected return lebih tinggi dari portofolio secara keseluruhan. Namun demikian, dari sisi risiko, portofolio menunjukkan efisiensi karena memiliki tingkat risiko yang lebih rendah dibandingkan dua saham penyusunnya. Estimasi risiko portofolio berdasarkan simulasi Monte Carlo menunjukkan nilai VaR sebesar 2,131852% dan CVaR sebesar 2,885113% pada tingkat kepercayaan 95%. Hasil backtesting menunjukkan bahwa estimasi risiko tersebut akurat dan dapat dijadikan acuan oleh investor dalam pengambilan keputusan investasi.
=====================================================================================================================================
The stock market is a promising investment instrument due to its high return potential, despite the significant risks involved. Therefore, accurate stock price forecasting and optimal portfolio construction are crucial for supporting better and more efficient investment decision-making. This study aims to (1) forecast the stock prices of IDX High Dividend 20 (IDXHIDIV20) using the hybrid ARIMA-LSTM method, (2) construct an optimal portfolio using the Single Index Model (SIM), and (3) estimate portfolio risk using Value at Risk (VaR) and Conditional Value at Risk (CVaR) through Monte Carlo simulation. The data used in this study consist of stocks consistently included in the IDXHIDIV20 index, namely ANTM, ASII, BBCA, BMRI, and TLKM, during the period from January 3, 2022 to January 3, 2025. The hybrid ARIMA-LSTM method is applied due to its ability to combine ARIMA's strength in capturing linear patterns and LSTM's capability in modeling complex non-linear patterns in time series data. Forecasting results show that the hybrid model yields very good prediction accuracy, with MAPE values ranging from 0.927293% to 3.702155%, influenced by the number of neurons in the LSTM architecture. Furthermore, the optimal SIM portfolio comprises three stocks: BBCA (32.43%) and BMRI (67.57%). This portfolio demonstrates a competitive expected return, although one of its components, BMRI, has a higher expected return than the overall portfolio. Nevertheless, the portfolio shows risk efficiency as it has a lower risk level compared to its constituent stocks. Portfolio risk estimation based on Monte Carlo simulation indicates a VaR of 2.131852% and a CVaR of 2.885113% at a 95% confidence level. Backtesting results confirm that these risk estimates are accurate and can serve as a reference for investors in making informed investment decisions.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	ARIMA-LSTM, IDXHIDIV20, Monte Carlo, Single Index Model, Value at Risk
Subjects:	Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA278.2 Regression Analysis. Logistic regression Q Science > QA Mathematics > QA280 Box-Jenkins forecasting Q Science > QA Mathematics > QA401 Mathematical models. Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) Q Science > QA Mathematics > QA76.9.D343 Data mining. Querying (Computer science)
Divisions:	Faculty of Mathematics, Computation, and Data Science > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User:	Angelica Benedict Setyadi
Date Deposited:	08 Aug 2025 01:20
Last Modified:	08 Aug 2025 01:20
URI:	http://repository.its.ac.id/id/eprint/124527

Actions (login required)

View Item