This could actually be really profitable. if you use machine learning to get the statistics of male female roles, their actors, how they acted and an array of other oscar winning qualities you can essentially come up with an oscar-winning algorithm
say the top 10 films in history have a 3.2:1 male to female ratio and the worst films have a 1:3 male female ratio. then you can gather that movies with more male roles statistically do better. boom, now you just figured out a hidden trend that can help your movie profit over others
-----
take good will hunting for example. mostly male roles. all protagonists are male. movie includes matt damon, about a student, etc. take all those together and make another mostly male movie with matt damon in it and you might just get another money making blockbuster