Implicit conversion from float -11.142857142857139 to int loses precision