Implicit conversion from float -40.07462686567163 to int loses precision