Implicit conversion from float -9.11602209944752 to int loses precision