Implicit conversion from float -75.30612244897961 to int loses precision