Implicit conversion from float -81.05769230769229 to int loses precision