Comparison of ARIMA and LSTM model on stock price predictions

Comparison of ARIMA and LSTM model on stock price predictions

Mathematics

Ka Ngai Wong

2025

Dr Salimeh Pour Mohammad

This research investigates and compares the predictive performance of two time-series forecasting methods—Auto-Regressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM) neural network for stock price prediction. The study uses historical stock price data obtained from Yahoo Finance to evaluate each model’s ability to predict trends. The ARIMA model, implemented using the “statsmodels” library, serves as a traditional statistical benchmark for modelling autocorrelated and stationary data. In contrast, the LSTM model, developed using “TensorFlow/Keras”, represents a modern deep learning approach capable of learning long-term dependencies and complex temporal dynamics. Data were divided into training and testing sets, with the final 15% reserved for out-of-sample evaluation. Model performance was assessed using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE).

machine learning, ARIMA, LSTM, stock prediction