Analysis and Prediction of Indonesia Stock Exchange (IDX) Stock Prices Using Long Short Term Memory (LSTM) Algorithm

Abstract

Stock investment is now a popular choice for many individuals and business entities. To optimize profits in investing, a deep understanding of price movements, timing, and accurate predictions in trading is required. The Long Short-Term Memory (LSTM) algorithm, a type of neural network suitable for time series data such as stock prices, can recognize complex temporal patterns in financial data. This algorithm has the potential to help investors and financial analysts predict BBRI stock prices more accurately. The purpose of this research is to predict the closing price of BBRI stock using the LSTM algorithm. This system can also conduct technical analysis with various indicators to understand the characteristics of the financial market. The research data includes BBRI stock prices from January 2006 to the present, with closing prices as the main variable. The research results show good model performance with a Mean Squared Error (MSE) of 0.000279, Mean Absolute Error (MAE) of 0.0133, and Root Mean Squared Error (RMSE) of 0.0167 on the training data. This reflects the model's level of accuracy against the training data. Although there is a slight increase in the validation data, these values remain within an acceptable level, indicating the model's ability to recognize data patterns that have not been seen before.