Implicit conversion from float -16.36363636363636 to int loses precision