Implicit conversion from float -17.611336032388664 to int loses precision