Implicit conversion from float -9.193548387096769 to int loses precision