Evaluating Models for Oil Price Forecasting

The present study is aimed at forecasting oil prices through the application of several quantitative models—moving averages, trend projection, Monte Carlo simulations, and autoregressive models. The analysis covers the period from 12 May 2024 to 12 May 2025, using daily-frequency historical data on Brent crude oil prices. For the moving average model, the classical accuracy indicators—Mean Absolute Deviation (MAD), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE)—were calculated and compared. In the case of trend projection, the evaluation was carried out using standard error, relative standard error, and Absolute Percentage Error (APE). For the Monte Carlo simulations and autoregressive models, the emphasis was placed on the process parameters, forecasted values, and confidence intervals, without using the full set of classical error indicators. The obtained results show that moving averages with shorter periods provide lower error, whereas longer periods lead to smoothing but also to a loss of accuracy. Trend projection and Monte Carlo simulations complement the analysis with a broader scenario-based perspective, while autoregressive models demonstrate good predictive performance in short-term forecasting. This highlights the importance of a combined approach in modelling oil price dynamics and provides a basis for more reliable forecasts under conditions of market uncertainty.