Implicit conversion from float -14.796663190823764 to int loses precision