The prices of shares prices are a critical task in financial analysis and investment decision - making. With the rise of strategies based on data based on statistical and machine learning, they have become the necessary tools for financial prognosis. This article represents a comparative study between ARIMA (Auto Regressive Integrated Moving Average), a traditional model of time series and LSTM (Long Short Term Model), a deep learning model, in predicting Tata Steel. Using historical data, we evaluate the performance of both models based on metrics such as Mae, RMSE and MAP. The results suggest that LSTM offers better performance when capturing non -linear patterns and dynamic market behavior.