Implicit conversion from float -72.56983240223462 to int loses precision