In this research, a method is proposed for predicting stock prices using deep learning techniques, specifically the Multivariate Sequential Long Short-Term Memory Autoencoder. This variant of the LSTM neural network model is designed to handle multivariate time series data. This model is able to capture dependencies between variables using the LSTM component, while the autoencoder component is used to learn a useful representation of the data in an unsupervised manner, which can improve the accuracy of stock price predictions. Its sequential structure also allows it to capture temporal dependencies in the data, making it well-suited for time series prediction tasks. The goal of this research is to help investors maximize returns through the identification of stock price trends. This paper also discusses the literature on various time series prediction models and describes the implementation and comparison of several techniques, including Univariate Sequential LSTM (USLSTM), Univariate Sequential LSTM Autoencoder (USLSTMA), Gated Recurrent Unit (GRU), Random Forest, and Generative Adversarial Networks (GAN). These models are compared with the proposed variants of Multivariate Sequential LSTM (MSLSTM), Multivariate Sequential LSTM Autoencoder (MSLSTMA) for market price prediction. In the experiments performed with real stock market data, the MSLSTMA model outperformed all other models in predicting the stock prices.