Implicit conversion from float -11.42857142857143 to int loses precision