Implicit conversion from float -22.70053475935829 to int loses precision