Implicit conversion from float -19.85130111524164 to int loses precision