An Extensive Statistical Analysis of Time Series Modeling and Forecasting of Crude Oil Prices

Crude oil is an important natural resource that is used to make fuels like jet fuel, petrol, and diesel, which are needed for industrial and transportation activities all over the world. It is essential to contemporary economies since it is used to produce a wide range of petrochemicals, which are used to make plastics, medications, and other common goods. The market of crude oil is volatile, and the price changes dramatically. As it is essential for daily activities of human life, many long-term and short-term decisions and work budget depend on the market price of this oil. Only forecasting accurately can provide insightful knowledge that helps to make the decisions and budgets more realistic. In this study, we analyze the time series crude oil price and forecast the oil price using different statistical methods. We employ Autoregressive (AR), Moving Average (MA), Autoregressive Integrated Moving Average (ARIMA), Seasonal Autoregressive Integrated Moving Average (SARIMA), Exponential Smoothing (ES), and Vector Autoregression (VAR) for analyzing and forecasting the oil price. We collect the daily Brent crude oil price data from marketwatch.com from May 20, 1987 to June 05, 2024. After handling the missing value, we check the stationarity of the dataset using Augmented Dickey-Fuller (ADF) Test and using differencing make the data more suitable to fit the statistical models. We calculate the models coefficients, statistics, autocorrelation function, partial autocorrelation function, simple moving average, exponential moving average, and residual analysis to get the insight of the crude oil price. Among all the models, VAR shows its superiority. This research will help the stockholders and researchers to take more suitable decision in different real-world scenarios.