Implicit conversion from float -18.5 to int loses precision