Implicit conversion from float -18.231707317073173 to int loses precision