Implicit conversion from float -14.014598540145997 to int loses precision