Implicit conversion from float -14.513888888888886 to int loses precision