Grid Search Algorithm Used for Optimizing Neural Networks in Stock Market Prediction

Predicting stock market trends and directions still attracts researchers, traders, and investors to investigate potential ways to secure their investments. In recent years, the complexity of the nature of the stock market has increased, creating a barrier between participants and their potential profits. In this study, we investigate the use of three different Machine Learning (ML) models: Logistic Regression (LR), Random Forest (RF), and Artificial Neural Network (ANN). Furthermore, Grid Search Optimization (GSO) will be used to tune the hyperparameters of the best model, among others, to maximize the accuracy prediction. The dataset contains historical stock prices spanning ten years from 2010 to 2020 and is pre-processed carefully to ensure data integrity. The results are assessed using Precision, Recall, F1-score, and Accuracy. Based on the results, using the GSO algorithm improved the accuracy of the ANN model by 10% compared with the default version.