IMPLEMENTATION OF LONG SHORT TERM MEMORY (LSTM) ALGORITHM FOR PREDICTING STOCK PRICE MOVEMENTS OF LQ45 INDEX (CASE STUDY: BBCA 2017 – 2023 STOCK PRICE)

Abstract

This research aims to implement the Long Short Term Memory (LSTM) algorithm in predicting the movement of LQ45 stock prices. In this study, historical data of BBCA stock prices were used as an example of LSTM method implementation. The development process of the stock price prediction application begins with the collection of historical data, which then undergoes a preprocessing stage for normalization. The data is divided into training and testing sets, and transformed into suitable sequences for LSTM model input. The LSTM model is trained using the backpropagation through time algorithm and tested using the testing data. The predicted results from the LSTM model are compared with the actual labels using RMSE and MAPE metrics. Once satisfactory predictions are obtained, they are stored in a database and presented to users in the form of graphs and comparison tables. The implementation of LSTM in this research demonstrates prediction accuracy with an error percentage below 6%, with MAPE of 5.4772% and RMSE of 6.658%. Furthermore, the implementation of LSTM in the developed application using the latest historical data also yields low error percentages, with MAPE ranging from 3.7763% to 5.8048% for various stock price features. In conclusion, the LSTM method can be used for predicting stock price movements with satisfactory accuracy, providing valuable information for investment decision-making.