Sample size for a study to validate a predictive model

how do i calculate the sample size for a validation of a prediction model. the rule of thumb i use is 10 events per variable. is there any better options?

In order to fit a reliable prognostic model, between 10 and 20 events per covariable are usually required, although this depends on the complexity of the model. More than 100 events are usually required for external validation, although this depends on the context.