Implicit conversion from float -27.02290076335879 to int loses precision