Predicting the stock market is always a difficult task due to its huge impact and dynamics. Traditional methods often fail to encapsulate the complex patterns and trends present in financial data. However, recent advancements in machine learning techniques have shown promising results in predicting stock market movements with higher accuracy and efficiency. This paper presents a comprehensive approach to securities market forecasting using machine learning (ML) algorithms. Traditional methods often face challenges in capturing the nonlinear and temporal dependencies present in financial time series data. In recent years, long short-term memory (LSTM) networks, a type of recurrent neural network (RNN), have gained fame for their ability to model continuous data effectively. This paper proposes a novel approach for stock market forecasting utilizing LSTM-based machine learning models. The methodology involves preprocessing historical stock market data, including price movements, trading volumes, and other respective indicators. Subsequently, the data is transformed into sequences suitable for training LSTM networks, preserving temporal dependencies and patterns. The LSTM models are trained using past years data to learn complex relationships and patterns inherent in the stock market. Numerous hyperparameters tuning strategies and regularization methods are employed to decorate the overall performance and the huge ability of the models.