The article proposes a new approach to predicting dynamic processes presented as a time series based on the Z number theory. Time series forecasting is a common problem in many real-life and science areas. An analysis of statistical, technical, and fundamental approaches to time series forecasting shows several shortcomings. This is primarily due to the lack of statistical data, the inability to use the knowledge and experience of specialists, the limited use of probabilistic methods, etc. Eliminating these shortcomings involves using fuzzy time series (FTS) and their extension – Z number-based time series and soft calculation methods. A major benefit of the fuzzy approach to forecasting is its capacity to include expert information or use a knowledge base created from data utilizing Data Mining technologies. This approach differs significantly from traditional methods by utilizing fuzzy time series, representing the variables using linguistic terms. Data is subjectively evaluated in this system, and the degree of confidence in the uncertainty is not expressed. A new Z-number-based method is proposed to bridge this gap. As an example this method is applied to forecast oil prices in the world market, demonstrating its practical applicability. As a strong foundation for the trustworthiness and effectiveness of the suggested technique, the experimental findings clearly show that the proposed forecasting method is more successful than current classical approaches.