Implicit conversion from float -4.261119081779057 to int loses precision