Forecasting oil prices is a crucial and formidable task due to its far-reaching implications on both economic and non-economic factors. The influence of various factors such as economic growth, political developments, and psychological sentiments adds significant uncertainty to oil price forecasts. Despite extensive research, there remains a lack of consensus among scholars regarding the most effective techniques and models for predicting oil prices. Therefore, it is imperative to develop forecasting methods that offer higher accuracy and lower error rates to address this challenge effectively. In this study, the OPEC basket price data from 01-2003 to 06-2022 was used to forecast oil prices using three forecasting methodologies: the Neural Network Autoregressive (NNAR) model (Soft-computing model), Fuzzy Time Series (FTS) (qualitative models), and the Prophet by Facebook model. The performance of the models in modeling and forecasting was evaluated by: Mean Error (ME), Root Mean Square Error (RMSE), mean absolute error (MAE), Mean Absolute Percentage Error (MAPE). The results of the study showed that the best model according to the previous criteria is Singh's model with: ME=0.268,MAE=2.801,RMSE=3.347,MAPE=4.708.\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ME=0.268, MAE=2.801, RMSE=3.347, MAPE=4.708.$$\end{document} the study contributes valuable insights to the field of oil price forecasting, paving the way for the adoption and implementation of Fuzzy Time Series models, particularly the Singh model, as indispensable tools in understanding and anticipating the dynamic behavior of oil markets.
