Implicit conversion from float -69.80769230769232 to int loses precision