Implicit conversion from float -26.564885496183223 to int loses precision