Estimasi Risiko Portofolio Optimal Model Mean Absolute Deviation Menggunakan Monte Carlo Berdasarkan Hasil Peramalan dengan Bidirectional Long Short-Term Memory

Fadillah, Adelia Yusrina (2025) Estimasi Risiko Portofolio Optimal Model Mean Absolute Deviation Menggunakan Monte Carlo Berdasarkan Hasil Peramalan dengan Bidirectional Long Short-Term Memory. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5006211006-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (6MB) | Request a copy

Abstract

Pasar modal Indonesia terus berkembang, dengan jumlah investor mencapai 12.326.700 per Januari 2024. Salah satu instrumen investasi yang banyak diminati adalah saham. Indeks harga saham yang dirilis oleh Bursa Efek Indonesia, salah satunya adalah IDXBUMN20 dengan kinerja yang unggul sejak tahun 2020. Selain itu, BUMN berkontribusi 20% dari total penerimaan negara dan memiliki pendapatan Rp 2.933 triliun pada 2023. Dalam berinvestasi, investor perlu mempertimbangkan return dan risiko yang dapat dilakukan melalui peramalan harga saham dan pembentukan portofolio optimal. Penelitian ini mengestimasi risiko portofolio saham optimal model Mean Absolute Deviation (MAD) dengan Monte Carlo berdasarkan data aktual dan hasil peramalan Bidirectional Long Short-Term Memory (Bi LSTM) selama 22 periode. Data yang digunakan adalah data closing price harian saham yang konsisten tergabung dalam indeks IDXBUMN20 selama periode 2 Agustus 2021 hingga 31 Juli 2024. Hasil penelitian menunjukkan bahwa nilai MAPE dari hasil prediksi saham menggunakan Bi-LSTM berada dalam rentang 0,927293% hingga 3,702155% yang dikategorikan sangat baik dipengaruhi banyaknya jumlah neuron. Kemudian, dibentuk portofolio optimal model MAD yang tersusun atas 8 saham, yaitu BBRI, BBNI, BBTN, ELSA, JSMR, PGAS, PTBA, dan TLKM. Estimasi tingkat risiko portofolio MAD berdasarkan Value at Risk (VaR) dan Conditional Value at Risk (CVaR) yang disimulasikan dengan Monte Carlo pada tingkat kepercayaan 95% masing-masing adalah 1,1101% dan 1,5602%. Pengujian backtesting VaR maupun CVaR akurat sehingga keduanya dapat digunakan investor sebagai acuan saat berinvestasi.
=====================================================================================================================================
Indonesian capital market continues to grow, with the number of investors reaching 12,326,700 as of January 2024. One of the most popular investment instruments is stocks. The stock price index released by the Indonesia Stock Exchange, one of which is IDXBUMN20 with superior performance since 2020. In addition, BUMN contributes 20% of total state revenue and has revenues of IDR 2,933 trillion in 2023. In investing, investors need to consider the returns and risks that can be done through stock price forecasting and optimal portfolio formation. This study estimates the risk of an optimal stock portfolio using the Mean Absolute Deviation (MAD) model with Monte Carlo based on actual data and Bidirectional Long Short-Term Memory (Bi LSTM) forecasting results for 22 periods. The data used is the daily closing price data of stocks that are consistently included in the IDXBUMN20 index during the period from August 2, 2021 to July 31, 2024. The results of the study show that the MAPE value of the stock prediction results using Bi-LSTM is in the range of 0,927293% to 3,702155% which is categorized as very good influenced by the number of neurons. Then, an optimal portfolio of the MAD model was formed consisting of 8 stocks, namely BBRI, BBNI, BBTN, ELSA, JSMR, PGAS, PTBA, and TLKM. The estimated risk level of the MAD portfolio based on Value at Risk (VaR) and Conditional Value at Risk (CVaR) simulated with Monte Carlo at a 95% confidence level is 1,1101% and 1,5602%, respectively. VaR and CVaR backtesting tests are accurate so that both can be used by investors as a reference when investing.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Bidirectional Long Short-Term Memory, Conditional Value at Risk, Mean Absolute Deviation, Monte Carlo, Value at Risk, Bidirectional Long Short-Term Memory
Subjects:	Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User:	Adelia Yusrina Fadillah
Date Deposited:	07 Jan 2025 08:51
Last Modified:	07 Jan 2025 08:51
URI:	http://repository.its.ac.id/id/eprint/116210

Actions (login required)

View Item