Stock Price Prediction on IDX30 Index using Long Short-Term Memory Algorithm

Abstract

The capital market plays a significant role in a country's economy, facilitating corporate financing and providing investment opportunities for the public. One popular investment instrument is stocks, yet many investors struggle to make profitable investment decisions due to a lack of understanding of stock investments. Therefore, predicting stock prices can be a way to determine the future value of a stock. This research aims to address this issue by applying the Long Short-Term Memory (LSTM) algorithm to predict stock prices on the IDX30 index. LSTM is capable of processing sequential data, such as stock price data, complexly because it can store information over long periods. The testing is conducted using various parameters in layers, epochs, and time steps to obtain the best prediction model. The LSTM architecture used consists of four layers: the LSTM layer with 128 neurons, dropout and dense layers with 64 neurons, and an additional dense layer that converts the output from the previous layer into prediction results. This study demonstrates that the LSTM algorithm can accurately predict stock prices based on evaluation metrics such as Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE). The best results for PT Bank Central Asia Tbk show a MAPE of 1.14% and RMSE of 137.71, PT Bank Rakyat Indonesia Tbk shows a MAPE of 1.58% and RMSE of 87.4, and PT Bank Mandiri Tbk shows a MAPE of 1.64% and RMSE of 88.26.

Keywords

• stocks prediction, stock prices, long short-term memory (lstm)