Implicit conversion from float -14.51612903225805 to int loses precision