TIME-SERIES ANALYSIS OF STOCK PRICES USING ARIMA AND SARIMA MODELS

Abstract

Stock price prediction is a great tool that helps investors to risk management and portfolio optimization. This research paper uses the Seasonal Auto Regressive Integrated Moving Average (SARIMA) model to forecast stock prices in the short term. It basically uses historical data to capture both the trend and seasonal patterns. The daily closing prices of companies which were saved in CSV files have been pre-processed to check their consistency and order in time. So the model will be able to make both retrospective (past ten days) and prospective (next ten days) predictions, which it will display in blue and red respectively, along with the actual prices for better understanding. The system built is in Python with Pandas, Matplotlib, and Stats models libraries; it is also deployed via a Streamlit interface. It effectively shows how SARIMA can spot short-term fluctuations and seasonal trends. As a result, they emphasize the importance of data preparation, model selection, and visualization in financial forecasting and open up the possibilities of future work in combining the external indicators or machine learning methods.