Implicit conversion from float -19.847560975609753 to int loses precision