Don't you need to account for the fact that the top 20% of attractive women in one context are not that attractive in another context. For example certain cultures value completely different features from others when it comes to attractiveness. In other cases, an individual might find multiple different features and ethnicity attractive such that they actually have a wideer pool of partners to select from. So unless your sample is completely homogenious and lacking significant cultural and ethnic diversity, I think the math would be far more complex than what you have.