Optimisasi Saham dan Reksa Dana Menggunakan Mean Absolute Deviation Berbasis Peramalan dengan Long Short-Term Memory

Yasa, Putu Dhiyo Devasya Sulindra (2026) Optimisasi Saham dan Reksa Dana Menggunakan Mean Absolute Deviation Berbasis Peramalan dengan Long Short-Term Memory. Other thesis, Insitut Teknologi Sepuluh Nopember.

Text 5006221080-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (6MB) | Request a copy

Abstract

Aktivitas investasi, di pasar modal Indonesia terus berkembang seiring meningkatnya partisipasi investor pada berbagai instrumen keuangan. Pengambilan keputusan investasi memerlukan strategi pengelolaan portofolio yang mampu menyeimbangkan return dan risiko, terutama pada aset dengan karakteristik berbeda seperti saham dan reksa dana. Penelitian ini mengintegrasikan metode Long Short-Term Memory (LSTM) untuk peramalan harga saham dan Nilai Aktiva Bersih (NAB) reksa dana, Mean Absolute Deviation (MAD) untuk pembentukan portofolio optimal, serta Value at Risk (VaR) dan Conditional Value at Risk (CVaR) untuk pengukuran risiko portofolio. Data yang digunakan meliputi harga penutupan harian saham BRPT, DEWA, dan ADMR, serta NAB harian reksa dana Schroder Dana Prestasi Plus, Manulife Dana Utama, dan Bahana Makara Abadi periode 2 Januari 2023 hingga 2 Januari 2026 dengan 713 observasi pada setiap aset. Model LSTM dibangun menggunakan pendekatan sliding window dengan time step 5 dan 10, kemudian dievaluasi menggunakan MAE, RMSE, dan MAPE. Hasil penelitian menunjukkan bahwa model LSTM memiliki kinerja peramalan yang baik, dengan nilai MAPE sebesar 0,101189% hingga 4,859294%. Optimisasi portofolio menggunakan MAD menghasilkan portofolio agresif yang didominasi saham ADMR, BRPT, dan DEWA dengan expected return tertinggi sebesar 0,001851, sedangkan portofolio moderat dan konservatif cenderung didominasi reksa dana Manulife dan Makara karena memiliki risiko lebih rendah dan return lebih stabil. Estimasi risiko, pada tingkat kepercayaan 95%, menghasilkan VaR sebesar 0,002066 hingga 0,034231 dan CVaR sebesar 0,004359 hingga 0,047001. Hasil backtesting menunjukkan estimasi risiko yang akurat, sehingga integrasi LSTM, MAD, VaR, dan CVaR dapat digunakan sebagai kerangka analisis investasi berbasis data dalam pembentukan portofolio saham dan reksa dana.
=====================================================================================================================================
Investment activity in the Indonesian capital market continues to grow, along with increasing investor participation in various financial instruments. Investment decision-making requires a portfolio management strategy capable of balancing return and risk, particularly for assets with different characteristics, such as stocks and mutual funds. This study integrates Long Short-Term Memory (LSTM) for forecasting stock prices and mutual fund Net Asset Value (NAV), Mean Absolute Deviation (MAD) for optimal portfolio construction, and Value at Risk (VaR) and Conditional Value at Risk (CVaR) for portfolio risk measurement. The data consist of the daily closing prices of BRPT, DEWA, and ADMR stocks, as well as the daily NAV of Schroder Dana Prestasi Plus, Manulife Dana Utama, and Bahana Makara Abadi mutual funds from 2 January 2023 to 2 January 2026, with 713 observations for each asset. The LSTM model was developed using a sliding window approach with time steps of 5 and 10, and was evaluated using MAE, RMSE, and MAPE. The results indicate that the LSTM model achieved good forecasting performance, with MAPE values ranging from 0.101189% to 4.859294%. Portfolio optimization using MAD produced an aggressive portfolio dominated by ADMR, BRPT, and DEWA stocks, yielding the highest expected return of 0.001851, whereas the moderate and conservative portfolios were primarily dominated by Manulife and Makara mutual funds due to their lower risk and more stable returns. The estimated risk at the 95% confidence level resulted in VaR values ranging from 0.002066 to 0.034231 and CVaR values ranging from 0.004359 to 0.047001. The backtesting results demonstrate accurate risk estimation, indicating that the integration of LSTM, MAD, VaR, and CVaR can serve as a data-driven investment analysis framework for constructing stock and mutual fund portfolios.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Conditional Value at Risk, Long Short-Term Memory, Mean Absolute Deviation, optimisasi portofolio, Value at Risk, Conditional Value at Risk, portfolio optimization
Subjects:	H Social Sciences > HA Statistics > HA30.3 Time-series analysis H Social Sciences > HD Industries. Land use. Labor > HD61 Risk Management H Social Sciences > HG Finance > HG4529 Investment analysis H Social Sciences > HG Finance > HG4529.5 Portfolio management H Social Sciences > HG Finance > HG4910 Investments H Social Sciences > HG Finance > HG4915 Stocks--Prices 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:	Putu Dhiyo Devasya Sulindra Yasa
Date Deposited:	17 Jul 2026 07:11
Last Modified:	17 Jul 2026 07:11
URI:	http://repository.its.ac.id/id/eprint/135338

Actions (login required)

View Item