Implicit conversion from float -15.0420168067227 to int loses precision