Stock market analysis and prediction is one of the most difficult tasks due to the high complexity and unpredictable nature of the stock market. The stock market operates within a broader ecosystem of interconnected markets, including commodities, currencies, and bonds, further amplifying its complexity. These magnify the challenge of predicting market movements accurately. Due to their capacity to process and analyze huge chunks of data in realtime, algorithms based on machine learning have become an effective tool for predicting future stock values. In this work, a method has been suggested to predict stock price and it is predicted whether to buy or sell a particular stock by considering the current stock market trend. A bunch of historical stock market data like open price, close price, high price and low price has been used as input to analyze and predict future closing prices. Long short-term memory (LSTM) network such as depth gated LSTM and peephole LSTM has been explored for predicting the stock price. Also, decision tree (DT) and k-nearest neighbors (KNN) for detecting the stock market trends has been explored. The lowest mean absolute error (MAE) is found to be 0.017 with peephole LSTM method for stock price prediction. The result produced by each algorithm has been compared with other existing methods. On the basis of the accuracy, effectiveness and closeness of the prediction, the best possible method for the stock prediction has been decided. The goal of the work is to help investors and financial experts make smarter and effective decisions by giving them insights about what might happen in the future.