Implicit conversion from float -18.05084745762713 to int loses precision