Implicit conversion from float -22.941176470588246 to int loses precision