Implicit conversion from float -52.27722772277227 to int loses precision