Implicit conversion from float -10.574712643678168 to int loses precision