Implicit conversion from float -52.30769230769229 to int loses precision