Fold-stratified cross-validation is a new validation methodology presented in the paper "Fold-stratified cross-validation for unbiased and privacy-preserving federated learning". It enables the unbiased validation of a model that has been trained on a multi-centric dataset using federated learning, while avoiding the retrieval of personally identifying information. In this jupyter notebook, Monte Carlo simulations are run to study the properties of stratified cross-validation using synthetic and real datasets.