Implicit conversion from float -9.197860962566835 to int loses precision