Hybrid Autoregressive Integrated Moving Average - Long Short-Term Memory (ARIMA-LSTM) Model for Forecasting BRI Bank Stock Prices

Dian Kurniasari (Universitas Lampung)
Silvi Fitriani (Universitas Lampung)
Warsono Warsono (Universitas Lampung)
Mustofa Usman (Universitas Lampung)

Jurnal Matematika UNAND
Vol. 15 No. 3 (2026)

Investors highly favour the BRI stock as one of the most popular investment options. The ARIMA method is a suitable approach for predicting stock prices. ARIMA is flawed because it assumes linearity. An alternative approach for predicting stock prices is using Long Short-Term Memory (LSTM) models. LSTM has advantages over ARIMA in handling both linear and nonlinear data. This study utilizes the ARIMA-LSTM hybrid approach to forecast the stock price of BRI. Utilizing a hybrid approach that involves dividing the data 90 percent of the data used for training The hybrid model, which uses 80 percent of the data for training and 20 percent for testing, does not achieve the same degree of accuracy as the model that uses 10 percent of the data for testing.