Implicit conversion from float -11.867219917012449 to int loses precision