Implicit conversion from float -57.69230769230771 to int loses precision