Implicit conversion from float -52.028985507246375 to int loses precision