Implicit conversion from float -9.097472924187727 to int loses precision