Implicit conversion from float -43.76623376623377 to int loses precision