Implicit conversion from float -5.029239766081872 to int loses precision