Card-Not-Present fraud uses the payment card information of a victim to buy in e-commerce platforms and later shows in the form of chargebacks. In 2024, it is expected to reach losses in the United States of 10 billion dollars. In the state of the art, the IEEE-CIS dataset has emerged as a strong option for creating smart detection systems against this problem. In this work, we create a solution that we call Hyphatia, where the novel Self-Supervised Learning paradigm is implemented in the tabular data domain using SubTab, outperforming XGBoost by 2.14% AUROC, detecting 67.44% of the fraud cases in the IEEE-CIS. This pioneering experimentation prioritizes those features that are not obfuscated, and beyond providing just classification metrics, we also provide time performance and feature importance calculations for explainability. To the best of our knowledge, this is one of the first works in the literature using the Self-Supervised Tabular Learning approach for the problem of credit card fraud detection.