Implicit conversion from float -30.876288659793815 to int loses precision