Implicit conversion from float -112.30769230769229 to int loses precision