Implicit conversion from float -13.548387096774178 to int loses precision