Implicit conversion from float -52.553191489361694 to int loses precision