Implicit conversion from float -11.123595505617985 to int loses precision