Implicit conversion from float -4.97206703910615 to int loses precision