Implicit conversion from float -133.95023328149298 to int loses precision