Implicit conversion from float -9.016853932584269 to int loses precision