Implicit conversion from float -26.593886462882097 to int loses precision