Implicit conversion from float -9.421487603305792 to int loses precision