Implicit conversion from float -13.96039603960395 to int loses precision