Implicit conversion from float -16.642335766423358 to int loses precision