Technical Indicator-Driven Stock Prediction: A LightGBM and ARIMA Perspective

The study examined the ability of a Light Gradient Boosting Machine (LightGBM) to predict stock prices with Exponential Moving Averages (EMA_5) and Simple Moving Averages (SMA_5) as primary Technical Indicators. Due to the autocorrelation in time series data, the adjusted closing prices were considered to run the model with several indicators effectively. These adjusted prices are subsequently used to compute the EMA_5 and SMA_5 as features in the model, to smooth price predictions and provide a clearer understanding of short-term market movements by identifying emerging patterns and trends with greater precision. The performance of LightBGMs will also be evaluated against ARIMA using statistical measures, including Coefficient of Determination (R2), Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE). The results indicated that for most instances, LightGBM outperforms ARIMA, producing R2 scores higher than 99%. It showcases the possibility of LightGBM as a robust stock market prediction model when it comes to the apparent complexities of market data. Future work will further extend the model's features to improve its already stunning, reliable prediction accuracy.