Predicting Stock Prices Using Deep Learning Models: CNN, LSTM, and CNN-LSTM. A Case Study of NVIDIA Stock Price Prediction for the Period 2020-2024

Abstract

This paper addresses the significance of stock price prediction for investors and the growing application of deep learning techniques, particularly advanced Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks, in forecasting stock prices. We propose a hybrid model, referred to as the Long Short-Term Memory-Convolutional Neural Network (LSTM-CNN) model, which integrates features extracted from different representations of the same dataset, specifically stock time series, to predict stock prices. The proposed model employs a CNN to extract features from the data and an LSTM network to retain temporal information and capture complex relationships between time points. The performance of the proposed model is evaluated against individual models and the hybrid model (CNN, LSTM, and LSTM-CNN) using data from Nvidia during the period from 01/01/2020 to 01/08/2024. Our LSTM-CNN feature integration model demonstrates superior performance in predicting stock prices compared to the individual models. Consequently, this study reveals that prediction errors can be efficiently minimized by combining both temporal and local features from the same dataset, rather than utilizing these features independently.