<p>This study develops a forecasting model for point time series, offering significant contributions at key phases. The first contribution involves constructing a new series that represents the percentage changes between consecutive points in the original series, and then grouping this new series into clusters with the suitable numbers, using cluster analysis technique. The second contribution quantifies the fuzzy relationship between future changes and historical data through the application of suitable rules. Finally, a new forecasting principle is established based on the above two improvements. The algorithm’s steps are described in detail, supported by a numerical example, and implemented using a Matlab program for real-time series analysis. The model’s effectiveness is validated through its superior performance compared to existing models on widely used benchmark series. Furthermore, the proposed model is applied to the stock market in Vietnam, achieving favorable forecasting results.</p>