Implicit conversion from float -14.571428571428555 to int loses precision