Implicit conversion from float -83.10344827586206 to int loses precision