Implicit conversion from float -11.847133757961785 to int loses precision