Implicit conversion from float -9.633027522935777 to int loses precision