Number of eras to embargo between train and test sets?

Just want to double check that I’m thinking about this correctly. In order to eliminate data leakage we should be using an embargo period of 4 eras for the 20 day targets and 12 eras for the 60 day targets?