Implicit conversion from float -88.57142857142856 to int loses precision