Implicit conversion from float -16.363636363636374 to int loses precision