Interval Forecasting of Stock Price Movements with Technical Indicator Input using Hybrid LASSO-Quantile Regression and Support Vector Regression

Forecasting stock prices is challenging due to the complex and nonlinear nature of financial markets. This study improves short- and medium-term forecasts for four major Indonesian banking stocks—BBCA, BBNI, BBRI, and BMRI—using a hybrid modeling approach. Daily stock data from September 2018 to December 2024 are used to calculate various technical indicators. LASSO-Quantile Regression selects the most important indicators, which are then applied to a Support Vector Regression (SVR) model to forecast price changes over 5-day and 22-day horizons. Model performance is evaluated through SMAPE over expanding-windows. Results show high forecast accuracy, with SMAPE values not exceed to 2%. The LASSO-QR model provides strong performance for short-term forecasts, while the hybrid model offers slight improvements for longer-term forecasts, particularly for BBCA and BBRI. Different technical indicators contribute depending on the time horizon—Comodity Chanel Index and Stochastic Oscillator enhance short-term accuracy, while Relative Strength Index and Moving Average Convergence Divergence are more useful for longer-term forecasts. Overall, the hybrid approach offers a reliable and efficient tool for forecasting stock prices in Indonesia’s banking sector.