Implicit conversion from float -9.559322033898297 to int loses precision