Peramalan Interval Arah Pergerakan Harga Saham dengan Input Indikator Teknikal Menggunakan Hybrid LASSO-Quantile Regression dan Support Vector Regression (LASSO-QR-SVR)

Rufida, Angger Salsabila (2025) Peramalan Interval Arah Pergerakan Harga Saham dengan Input Indikator Teknikal Menggunakan Hybrid LASSO-Quantile Regression dan Support Vector Regression (LASSO-QR-SVR). Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6003232013-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Peramalan arah pergerakan harga saham menjadi aspek penting dalam pengambilan keputusan investasi, khususnya di sektor perbankan yang memiliki peran strategis dan tingkat likuiditas tinggi dalam perekonomian Indonesia. Penelitian ini bertujuan untuk meramalkan perubahan harga saham pada perusahaan perbankan besar di Indonesia—yaitu Bank Central Asia (BBCA), Bank Negara Indonesia (BBNI), Bank Rakyat Indonesia (BBRI), dan Bank Mandiri (BMRI)—dengan memanfaatkan indikator teknikal sebagai variabel prediktor. Model yang digunakan adalah pendekatan hybrid LASSO-Quantile Regression (LASSO-QR) dan Support Vector Regression (SVR), dengan membagi data menggunakan kerangka expanding window serta tiga horizon waktu prediksi (5, 10, dan 22 hari). LASSO-QR digunakan untuk seleksi fitur dan penentuan kuantil prediksi, sedangkan SVR digunakan untuk menghasilkan prediksi perubahan harga berdasarkan variabel yang telah diseleksi tiap window. Hasil penelitian menunjukkan bahwa indikator Moving Average (MA) merupakan variabel teknikal yang paling konsisten terpilih di berbagai kuantil dan horizon, dengan tingkat terpilih tinggi berkisar antara 50% hingga 100%. Evaluasi kinerja model mencakup nilai RMSE, akurasi arah (Directional Accuracy) dan cakupan interval prediksi. Meskipun rata-rata DA hanya berada pada kisaran 43% hingga 49%, model menunjukkan kemampuan dalam menangkap data aktual dalam interval kuantil mencapai 72% hingga 88%. Selain itu, hasil juga menunjukkan bahwa semakin panjang horizon prediksi, semakin lebar pula rentang interval kuantil yang mencerminkan tingkat ketidakpastian pasar yang meningkat dalam jangka waktu yang lebih panjang. Secara keseluruhan, pendekatan hybrid LASSO-QR dan SVR mampu memberikan gambaran prediktif berupa rentang pergerakan harga saham, sehingga investor dapat mengidentifikasi kemungkinan perubahan harga terendah dan tertinggi di masa depan.
========================================================================================================================================
Forecasting the direction of stock price movements is a crucial aspect of investment decision-making, particularly in the banking sector, which plays a strategic role and exhibits high liquidity within Indonesia’s economy. This study aims to forecast stock price changes of major Indonesian banking companies—namely Bank Central Asia (BBCA), Bank Negara Indonesia (BBNI), Bank Rakyat Indonesia (BBRI), and Bank Mandiri (BMRI)—by utilizing technical indicators as predictor variables. The proposed model employs a hybrid approach combining LASSO-Quantile Regression (LASSO-QR) and Support Vector Regression (SVR), with an expanding window framework and three forecasting horizons (5, 10, and 22 days). LASSO-QR is used for feature selection and quantile estimation, while SVR generates point forecasts based on the selected variables for each window. The results show that the Moving Average (MA) indicator is the most consistently selected technical variable across quantiles and horizons, with a high selection rate ranging from 50% to 100%. Model performance is evaluated using RMSE, directional accuracy (DA), and prediction interval coverage. Although the average DA is relatively low, ranging from 43% to 49%, the model demonstrates strong interval coverage, capturing actual data within the predicted quantile ranges in 72% to 88% of cases. Furthermore, the findings indicate that longer forecasting horizons lead to wider quantile intervals, reflecting greater market uncertainty over extended periods. Overall, the hybrid LASSO-QR and SVR approach provides predictive insights in the form of stock price movement intervals, allowing investors to identify potential lower and upper bounds of future price changes for more informed technical indicator-based investment decisions.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Saham, Indikator teknikal, LASSO-QR, SVR, Stock, Technical indicators
Subjects:	H Social Sciences > HA Statistics > HA30.3 Time-series analysis H Social Sciences > HG Finance > HG4529 Investment analysis H Social Sciences > HG Finance > HG4910 Investments H Social Sciences > HG Finance > HG4915 Stocks--Prices
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis
Depositing User:	Angger Salsabila Rufida
Date Deposited:	04 Aug 2025 06:36
Last Modified:	04 Aug 2025 06:36
URI:	http://repository.its.ac.id/id/eprint/125482

Actions (login required)

View Item