Implicit conversion from float -23.59223300970875 to int loses precision