This paper addresses the challenge of credit card fraud detection in the context of continual learning, focusing on balancing plasticity (learning new patterns) and stability (remembering old ones). The authors propose an evaluation procedure, the plasticity/stability visualization matrix, to quantify the forgetting in data streams with delayed feedback. They compare six strategies and 13 different models on a real-world dataset of over 50 million e-commerce transactions. The study highlights the trade-off between plasticity and stability, concluding that the best method is an ensemble of frozen neural networks (FrzE), which performs well in terms of precision at 100 (Pr@100) and area under the precision-recall curve (AUPRC) while achieving satisfactory reduced forgetting. The findings suggest that while high Pr@100 is crucial for model selection, low forgetting should also be prioritized when similar Pr@100 performances are achieved. The research provides valuable insights for improving fraud detection systems in the fintech domain.This paper addresses the challenge of credit card fraud detection in the context of continual learning, focusing on balancing plasticity (learning new patterns) and stability (remembering old ones). The authors propose an evaluation procedure, the plasticity/stability visualization matrix, to quantify the forgetting in data streams with delayed feedback. They compare six strategies and 13 different models on a real-world dataset of over 50 million e-commerce transactions. The study highlights the trade-off between plasticity and stability, concluding that the best method is an ensemble of frozen neural networks (FrzE), which performs well in terms of precision at 100 (Pr@100) and area under the precision-recall curve (AUPRC) while achieving satisfactory reduced forgetting. The findings suggest that while high Pr@100 is crucial for model selection, low forgetting should also be prioritized when similar Pr@100 performances are achieved. The research provides valuable insights for improving fraud detection systems in the fintech domain.