Implicit conversion from float -22.2340425531915 to int loses precision