Software development effort estimation is strategic for organizations. In Agile methodologies, Story Points (SP) are commonly used to measure the effort required to complete a user story. While machine learning techniques have been explored for effort estimation, existing models, especially those using Deep Learning, can be time-consuming for feature extraction and training. This study addresses the research gap by proposing and validating a Story Point predictor. The model uses features extracted from user story titles and descriptions, specifically focusing on readability, sentiment, and subjectivity indicators. The study used an experimental methodology with a quantitative approach, validated on a dataset of 34 open-source projects. Evaluation was performed using the Mean Absolute Error (MAE) metric and the Wilcoxon test for statistical significance. Two baselines, Mean-Based Regression (MbR) and TFIDFSE, were used for comparison. The results show that the model achieved a better (lower) apparent MAE than MbR and TFIDFSE in most of the projects. Furthermore, the proposed approach for feature extraction and training is significantly faster than the baselines. Extracting readability, sentiment, and subjectivity features from user stories shows promise in improving estimation accuracy and efficiency.