Implicit conversion from float -75.16853932584269 to int loses precision