Implicit conversion from float -57.15517241379311 to int loses precision